From 8b78cf602fd3bd97c0080edd22fe8fd5d0fa7832 Mon Sep 17 00:00:00 2001 From: Yi Yang Date: Mon, 25 Feb 2008 08:46:12 +0800 Subject: cpuidle: fix cpuidle time and usage overflow cpuidle C-state sysfs node time and usage are very easy to overflow because they are all of unsigned int type, time will overflow within about two hours, usage will take longer time to overflow, but they are increasing for ever. This patch will convert them to unsigned long long. Signed-off-by: Yi Yang Acked-by: Venkatesh Pallipadi Signed-off-by: Len Brown --- include/linux/cpuidle.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/cpuidle.h b/include/linux/cpuidle.h index 6b72a4584086..51e6b1e520e6 100644 --- a/include/linux/cpuidle.h +++ b/include/linux/cpuidle.h @@ -38,8 +38,8 @@ struct cpuidle_state { unsigned int power_usage; /* in mW */ unsigned int target_residency; /* in US */ - unsigned int usage; - unsigned int time; /* in US */ + unsigned long long usage; + unsigned long long time; /* in US */ int (*enter) (struct cpuidle_device *dev, struct cpuidle_state *state); -- cgit v1.3 From 06d8308c61e54346585b2691c13ee3f90cb6fb2f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sat, 22 Mar 2008 09:20:24 +0100 Subject: NOHZ: reevaluate idle sleep length after add_timer_on() add_timer_on() can add a timer on a CPU which is currently in a long idle sleep, but the timer wheel is not reevaluated by the nohz code on that CPU. So a timer can be delayed for quite a long time. This triggered a false positive in the clocksource watchdog code. To avoid this we need to wake up the idle CPU and enforce the reevaluation of the timer wheel for the next timer event. Add a function, which checks a given CPU for idle state, marks the idle task with NEED_RESCHED and sends a reschedule IPI to notify the other CPU of the change in the timer wheel. Call this function from add_timer_on(). Signed-off-by: Thomas Gleixner Acked-by: Peter Zijlstra Acked-by: Ingo Molnar Cc: stable@kernel.org -- include/linux/sched.h | 6 ++++++ kernel/sched.c | 43 +++++++++++++++++++++++++++++++++++++++++++ kernel/timer.c | 10 +++++++++- 3 files changed, 58 insertions(+), 1 deletion(-) --- include/linux/sched.h | 6 ++++++ kernel/sched.c | 43 +++++++++++++++++++++++++++++++++++++++++++ kernel/timer.c | 10 +++++++++- 3 files changed, 58 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/sched.h b/include/linux/sched.h index fed07d03364e..6a1e7afb099b 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1541,6 +1541,12 @@ static inline void idle_task_exit(void) {} extern void sched_idle_next(void); +#if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) +extern void wake_up_idle_cpu(int cpu); +#else +static inline void wake_up_idle_cpu(int cpu) { } +#endif + #ifdef CONFIG_SCHED_DEBUG extern unsigned int sysctl_sched_latency; extern unsigned int sysctl_sched_min_granularity; diff --git a/kernel/sched.c b/kernel/sched.c index 28c73f07efb2..8dcdec6fe0fe 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1052,6 +1052,49 @@ static void resched_cpu(int cpu) resched_task(cpu_curr(cpu)); spin_unlock_irqrestore(&rq->lock, flags); } + +#ifdef CONFIG_NO_HZ +/* + * When add_timer_on() enqueues a timer into the timer wheel of an + * idle CPU then this timer might expire before the next timer event + * which is scheduled to wake up that CPU. In case of a completely + * idle system the next event might even be infinite time into the + * future. wake_up_idle_cpu() ensures that the CPU is woken up and + * leaves the inner idle loop so the newly added timer is taken into + * account when the CPU goes back to idle and evaluates the timer + * wheel for the next timer event. + */ +void wake_up_idle_cpu(int cpu) +{ + struct rq *rq = cpu_rq(cpu); + + if (cpu == smp_processor_id()) + return; + + /* + * This is safe, as this function is called with the timer + * wheel base lock of (cpu) held. When the CPU is on the way + * to idle and has not yet set rq->curr to idle then it will + * be serialized on the timer wheel base lock and take the new + * timer into account automatically. + */ + if (rq->curr != rq->idle) + return; + + /* + * We can set TIF_RESCHED on the idle task of the other CPU + * lockless. The worst case is that the other CPU runs the + * idle task through an additional NOOP schedule() + */ + set_tsk_thread_flag(rq->idle, TIF_NEED_RESCHED); + + /* NEED_RESCHED must be visible before we test polling */ + smp_mb(); + if (!tsk_is_polling(rq->idle)) + smp_send_reschedule(cpu); +} +#endif + #else static void __resched_task(struct task_struct *p, int tif_bit) { diff --git a/kernel/timer.c b/kernel/timer.c index 99b00a25f88b..b024106daa70 100644 --- a/kernel/timer.c +++ b/kernel/timer.c @@ -451,10 +451,18 @@ void add_timer_on(struct timer_list *timer, int cpu) spin_lock_irqsave(&base->lock, flags); timer_set_base(timer, base); internal_add_timer(base, timer); + /* + * Check whether the other CPU is idle and needs to be + * triggered to reevaluate the timer wheel when nohz is + * active. We are protected against the other CPU fiddling + * with the timer by holding the timer base lock. This also + * makes sure that a CPU on the way to idle can not evaluate + * the timer wheel. + */ + wake_up_idle_cpu(cpu); spin_unlock_irqrestore(&base->lock, flags); } - /** * mod_timer - modify a timer's timeout * @timer: the timer to be modified -- cgit v1.3 From a6bd8e13034dd7d60b6f14217096efa192d0adc1 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Fri, 28 Mar 2008 11:05:53 -0500 Subject: lguest: comment documentation update. Took some cycles to re-read the Lguest Journey end-to-end, fix some rot and tighten some phrases. Only comments change. No new jokes, but a couple of recycled old jokes. Signed-off-by: Rusty Russell --- Documentation/lguest/lguest.c | 69 ++++++++++++---------- arch/x86/lguest/boot.c | 108 +++++++++++++++++++--------------- arch/x86/lguest/i386_head.S | 15 +++-- drivers/lguest/core.c | 18 +++--- drivers/lguest/hypercalls.c | 11 +++- drivers/lguest/interrupts_and_traps.c | 7 +-- drivers/lguest/lguest_device.c | 11 ++-- drivers/lguest/lguest_user.c | 30 +++++++--- drivers/lguest/page_tables.c | 32 +++++----- drivers/lguest/x86/core.c | 33 +++++++---- drivers/lguest/x86/switcher_32.S | 8 +-- include/asm-x86/lguest_hcall.h | 2 +- include/linux/lguest_launcher.h | 6 +- 13 files changed, 208 insertions(+), 142 deletions(-) (limited to 'include/linux') diff --git a/Documentation/lguest/lguest.c b/Documentation/lguest/lguest.c index d45c7f682b1b..4c1fc65a8b3d 100644 --- a/Documentation/lguest/lguest.c +++ b/Documentation/lguest/lguest.c @@ -1,7 +1,7 @@ /*P:100 This is the Launcher code, a simple program which lays out the - * "physical" memory for the new Guest by mapping the kernel image and the - * virtual devices, then reads repeatedly from /dev/lguest to run the Guest. -:*/ + * "physical" memory for the new Guest by mapping the kernel image and + * the virtual devices, then opens /dev/lguest to tell the kernel + * about the Guest and control it. :*/ #define _LARGEFILE64_SOURCE #define _GNU_SOURCE #include @@ -43,7 +43,7 @@ #include "linux/virtio_console.h" #include "linux/virtio_ring.h" #include "asm-x86/bootparam.h" -/*L:110 We can ignore the 38 include files we need for this program, but I do +/*L:110 We can ignore the 39 include files we need for this program, but I do * want to draw attention to the use of kernel-style types. * * As Linus said, "C is a Spartan language, and so should your naming be." I @@ -320,7 +320,7 @@ static unsigned long map_elf(int elf_fd, const Elf32_Ehdr *ehdr) err(1, "Reading program headers"); /* Try all the headers: there are usually only three. A read-only one, - * a read-write one, and a "note" section which isn't loadable. */ + * a read-write one, and a "note" section which we don't load. */ for (i = 0; i < ehdr->e_phnum; i++) { /* If this isn't a loadable segment, we ignore it */ if (phdr[i].p_type != PT_LOAD) @@ -387,7 +387,7 @@ static unsigned long load_kernel(int fd) if (memcmp(hdr.e_ident, ELFMAG, SELFMAG) == 0) return map_elf(fd, &hdr); - /* Otherwise we assume it's a bzImage, and try to unpack it */ + /* Otherwise we assume it's a bzImage, and try to load it. */ return load_bzimage(fd); } @@ -433,12 +433,12 @@ static unsigned long load_initrd(const char *name, unsigned long mem) return len; } -/* Once we know how much memory we have, we can construct simple linear page +/* Once we know how much memory we have we can construct simple linear page * tables which set virtual == physical which will get the Guest far enough * into the boot to create its own. * * We lay them out of the way, just below the initrd (which is why we need to - * know its size). */ + * know its size here). */ static unsigned long setup_pagetables(unsigned long mem, unsigned long initrd_size) { @@ -850,7 +850,8 @@ static void handle_console_output(int fd, struct virtqueue *vq) * * Handling output for network is also simple: we get all the output buffers * and write them (ignoring the first element) to this device's file descriptor - * (stdout). */ + * (/dev/net/tun). + */ static void handle_net_output(int fd, struct virtqueue *vq) { unsigned int head, out, in; @@ -924,7 +925,7 @@ static void enable_fd(int fd, struct virtqueue *vq) write(waker_fd, &vq->dev->fd, sizeof(vq->dev->fd)); } -/* Resetting a device is fairly easy. */ +/* When the Guest asks us to reset a device, it's is fairly easy. */ static void reset_device(struct device *dev) { struct virtqueue *vq; @@ -1003,8 +1004,8 @@ static void handle_input(int fd) if (select(devices.max_infd+1, &fds, NULL, NULL, &poll) == 0) break; - /* Otherwise, call the device(s) which have readable - * file descriptors and a method of handling them. */ + /* Otherwise, call the device(s) which have readable file + * descriptors and a method of handling them. */ for (i = devices.dev; i; i = i->next) { if (i->handle_input && FD_ISSET(i->fd, &fds)) { int dev_fd; @@ -1015,8 +1016,7 @@ static void handle_input(int fd) * should no longer service it. Networking and * console do this when there's no input * buffers to deliver into. Console also uses - * it when it discovers that stdin is - * closed. */ + * it when it discovers that stdin is closed. */ FD_CLR(i->fd, &devices.infds); /* Tell waker to ignore it too, by sending a * negative fd number (-1, since 0 is a valid @@ -1033,7 +1033,8 @@ static void handle_input(int fd) * * All devices need a descriptor so the Guest knows it exists, and a "struct * device" so the Launcher can keep track of it. We have common helper - * routines to allocate and manage them. */ + * routines to allocate and manage them. + */ /* The layout of the device page is a "struct lguest_device_desc" followed by a * number of virtqueue descriptors, then two sets of feature bits, then an @@ -1078,7 +1079,7 @@ static void add_virtqueue(struct device *dev, unsigned int num_descs, struct virtqueue **i, *vq = malloc(sizeof(*vq)); void *p; - /* First we need some pages for this virtqueue. */ + /* First we need some memory for this virtqueue. */ pages = (vring_size(num_descs, getpagesize()) + getpagesize() - 1) / getpagesize(); p = get_pages(pages); @@ -1122,7 +1123,7 @@ static void add_virtqueue(struct device *dev, unsigned int num_descs, } /* The first half of the feature bitmask is for us to advertise features. The - * second half if for the Guest to accept features. */ + * second half is for the Guest to accept features. */ static void add_feature(struct device *dev, unsigned bit) { u8 *features = get_feature_bits(dev); @@ -1151,7 +1152,9 @@ static void set_config(struct device *dev, unsigned len, const void *conf) } /* This routine does all the creation and setup of a new device, including - * calling new_dev_desc() to allocate the descriptor and device memory. */ + * calling new_dev_desc() to allocate the descriptor and device memory. + * + * See what I mean about userspace being boring? */ static struct device *new_device(const char *name, u16 type, int fd, bool (*handle_input)(int, struct device *)) { @@ -1492,7 +1495,10 @@ static int io_thread(void *_dev) while (read(vblk->workpipe[0], &c, 1) == 1) { /* We acknowledge each request immediately to reduce latency, * rather than waiting until we've done them all. I haven't - * measured to see if it makes any difference. */ + * measured to see if it makes any difference. + * + * That would be an interesting test, wouldn't it? You could + * also try having more than one I/O thread. */ while (service_io(dev)) write(vblk->done_fd, &c, 1); } @@ -1500,7 +1506,7 @@ static int io_thread(void *_dev) } /* Now we've seen the I/O thread, we return to the Launcher to see what happens - * when the thread tells us it's completed some I/O. */ + * when that thread tells us it's completed some I/O. */ static bool handle_io_finish(int fd, struct device *dev) { char c; @@ -1572,11 +1578,12 @@ static void setup_block_file(const char *filename) * more work. */ pipe(vblk->workpipe); - /* Create stack for thread and run it */ + /* Create stack for thread and run it. Since stack grows upwards, we + * point the stack pointer to the end of this region. */ stack = malloc(32768); /* SIGCHLD - We dont "wait" for our cloned thread, so prevent it from * becoming a zombie. */ - if (clone(io_thread, stack + 32768, CLONE_VM | SIGCHLD, dev) == -1) + if (clone(io_thread, stack + 32768, CLONE_VM | SIGCHLD, dev) == -1) err(1, "Creating clone"); /* We don't need to keep the I/O thread's end of the pipes open. */ @@ -1586,14 +1593,14 @@ static void setup_block_file(const char *filename) verbose("device %u: virtblock %llu sectors\n", devices.device_num, le64_to_cpu(conf.capacity)); } -/* That's the end of device setup. :*/ +/* That's the end of device setup. */ -/* Reboot */ +/*L:230 Reboot is pretty easy: clean up and exec() the Launcher afresh. */ static void __attribute__((noreturn)) restart_guest(void) { unsigned int i; - /* Closing pipes causes the waker thread and io_threads to die, and + /* Closing pipes causes the Waker thread and io_threads to die, and * closing /dev/lguest cleans up the Guest. Since we don't track all * open fds, we simply close everything beyond stderr. */ for (i = 3; i < FD_SETSIZE; i++) @@ -1602,7 +1609,7 @@ static void __attribute__((noreturn)) restart_guest(void) err(1, "Could not exec %s", main_args[0]); } -/*L:220 Finally we reach the core of the Launcher, which runs the Guest, serves +/*L:220 Finally we reach the core of the Launcher which runs the Guest, serves * its input and output, and finally, lays it to rest. */ static void __attribute__((noreturn)) run_guest(int lguest_fd) { @@ -1643,7 +1650,7 @@ static void __attribute__((noreturn)) run_guest(int lguest_fd) err(1, "Resetting break"); } } -/* +/*L:240 * This is the end of the Launcher. The good news: we are over halfway * through! The bad news: the most fiendish part of the code still lies ahead * of us. @@ -1690,8 +1697,8 @@ int main(int argc, char *argv[]) * device receive input from a file descriptor, we keep an fdset * (infds) and the maximum fd number (max_infd) with the head of the * list. We also keep a pointer to the last device. Finally, we keep - * the next interrupt number to hand out (1: remember that 0 is used by - * the timer). */ + * the next interrupt number to use for devices (1: remember that 0 is + * used by the timer). */ FD_ZERO(&devices.infds); devices.max_infd = -1; devices.lastdev = NULL; @@ -1792,8 +1799,8 @@ int main(int argc, char *argv[]) lguest_fd = tell_kernel(pgdir, start); /* We fork off a child process, which wakes the Launcher whenever one - * of the input file descriptors needs attention. Otherwise we would - * run the Guest until it tries to output something. */ + * of the input file descriptors needs attention. We call this the + * Waker, and we'll cover it in a moment. */ waker_fd = setup_waker(lguest_fd); /* Finally, run the Guest. This doesn't return. */ diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c index a104c532ff70..3335b4595efd 100644 --- a/arch/x86/lguest/boot.c +++ b/arch/x86/lguest/boot.c @@ -10,21 +10,19 @@ * (such as the example in Documentation/lguest/lguest.c) is called the * Launcher. * - * Secondly, we only run specially modified Guests, not normal kernels. When - * you set CONFIG_LGUEST to 'y' or 'm', this automatically sets - * CONFIG_LGUEST_GUEST=y, which compiles this file into the kernel so it knows - * how to be a Guest. This means that you can use the same kernel you boot - * normally (ie. as a Host) as a Guest. + * Secondly, we only run specially modified Guests, not normal kernels: setting + * CONFIG_LGUEST_GUEST to "y" compiles this file into the kernel so it knows + * how to be a Guest at boot time. This means that you can use the same kernel + * you boot normally (ie. as a Host) as a Guest. * * These Guests know that they cannot do privileged operations, such as disable * interrupts, and that they have to ask the Host to do such things explicitly. * This file consists of all the replacements for such low-level native * hardware operations: these special Guest versions call the Host. * - * So how does the kernel know it's a Guest? The Guest starts at a special - * entry point marked with a magic string, which sets up a few things then - * calls here. We replace the native functions various "paravirt" structures - * with our Guest versions, then boot like normal. :*/ + * So how does the kernel know it's a Guest? We'll see that later, but let's + * just say that we end up here where we replace the native functions various + * "paravirt" structures with our Guest versions, then boot like normal. :*/ /* * Copyright (C) 2006, Rusty Russell IBM Corporation. @@ -134,7 +132,7 @@ static void async_hcall(unsigned long call, unsigned long arg1, * lguest_leave_lazy_mode(). * * So, when we're in lazy mode, we call async_hcall() to store the call for - * future processing. */ + * future processing: */ static void lazy_hcall(unsigned long call, unsigned long arg1, unsigned long arg2, @@ -147,7 +145,7 @@ static void lazy_hcall(unsigned long call, } /* When lazy mode is turned off reset the per-cpu lazy mode variable and then - * issue a hypercall to flush any stored calls. */ + * issue the do-nothing hypercall to flush any stored calls. */ static void lguest_leave_lazy_mode(void) { paravirt_leave_lazy(paravirt_get_lazy_mode()); @@ -164,7 +162,7 @@ static void lguest_leave_lazy_mode(void) * * So instead we keep an "irq_enabled" field inside our "struct lguest_data", * which the Guest can update with a single instruction. The Host knows to - * check there when it wants to deliver an interrupt. + * check there before it tries to deliver an interrupt. */ /* save_flags() is expected to return the processor state (ie. "flags"). The @@ -196,10 +194,15 @@ static void irq_enable(void) /*M:003 Note that we don't check for outstanding interrupts when we re-enable * them (or when we unmask an interrupt). This seems to work for the moment, * since interrupts are rare and we'll just get the interrupt on the next timer - * tick, but when we turn on CONFIG_NO_HZ, we should revisit this. One way + * tick, but now we can run with CONFIG_NO_HZ, we should revisit this. One way * would be to put the "irq_enabled" field in a page by itself, and have the * Host write-protect it when an interrupt comes in when irqs are disabled. - * There will then be a page fault as soon as interrupts are re-enabled. :*/ + * There will then be a page fault as soon as interrupts are re-enabled. + * + * A better method is to implement soft interrupt disable generally for x86: + * instead of disabling interrupts, we set a flag. If an interrupt does come + * in, we then disable them for real. This is uncommon, so we could simply use + * a hypercall for interrupt control and not worry about efficiency. :*/ /*G:034 * The Interrupt Descriptor Table (IDT). @@ -212,6 +215,10 @@ static void irq_enable(void) static void lguest_write_idt_entry(gate_desc *dt, int entrynum, const gate_desc *g) { + /* The gate_desc structure is 8 bytes long: we hand it to the Host in + * two 32-bit chunks. The whole 32-bit kernel used to hand descriptors + * around like this; typesafety wasn't a big concern in Linux's early + * years. */ u32 *desc = (u32 *)g; /* Keep the local copy up to date. */ native_write_idt_entry(dt, entrynum, g); @@ -243,7 +250,8 @@ static void lguest_load_idt(const struct desc_ptr *desc) * * This is the opposite of the IDT code where we have a LOAD_IDT_ENTRY * hypercall and use that repeatedly to load a new IDT. I don't think it - * really matters, but wouldn't it be nice if they were the same? + * really matters, but wouldn't it be nice if they were the same? Wouldn't + * it be even better if you were the one to send the patch to fix it? */ static void lguest_load_gdt(const struct desc_ptr *desc) { @@ -298,9 +306,9 @@ static void lguest_load_tr_desc(void) /* The "cpuid" instruction is a way of querying both the CPU identity * (manufacturer, model, etc) and its features. It was introduced before the - * Pentium in 1993 and keeps getting extended by both Intel and AMD. As you - * might imagine, after a decade and a half this treatment, it is now a giant - * ball of hair. Its entry in the current Intel manual runs to 28 pages. + * Pentium in 1993 and keeps getting extended by both Intel, AMD and others. + * As you might imagine, after a decade and a half this treatment, it is now a + * giant ball of hair. Its entry in the current Intel manual runs to 28 pages. * * This instruction even it has its own Wikipedia entry. The Wikipedia entry * has been translated into 4 languages. I am not making this up! @@ -594,17 +602,17 @@ static unsigned long lguest_get_wallclock(void) return lguest_data.time.tv_sec; } -/* The TSC is a Time Stamp Counter. The Host tells us what speed it runs at, - * or 0 if it's unusable as a reliable clock source. This matches what we want - * here: if we return 0 from this function, the x86 TSC clock will not register - * itself. */ +/* The TSC is an Intel thing called the Time Stamp Counter. The Host tells us + * what speed it runs at, or 0 if it's unusable as a reliable clock source. + * This matches what we want here: if we return 0 from this function, the x86 + * TSC clock will give up and not register itself. */ static unsigned long lguest_cpu_khz(void) { return lguest_data.tsc_khz; } -/* If we can't use the TSC, the kernel falls back to our "lguest_clock", where - * we read the time value given to us by the Host. */ +/* If we can't use the TSC, the kernel falls back to our lower-priority + * "lguest_clock", where we read the time value given to us by the Host. */ static cycle_t lguest_clock_read(void) { unsigned long sec, nsec; @@ -648,12 +656,16 @@ static struct clocksource lguest_clock = { static int lguest_clockevent_set_next_event(unsigned long delta, struct clock_event_device *evt) { + /* FIXME: I don't think this can ever happen, but James tells me he had + * to put this code in. Maybe we should remove it now. Anyone? */ if (delta < LG_CLOCK_MIN_DELTA) { if (printk_ratelimit()) printk(KERN_DEBUG "%s: small delta %lu ns\n", __FUNCTION__, delta); return -ETIME; } + + /* Please wake us this far in the future. */ hcall(LHCALL_SET_CLOCKEVENT, delta, 0, 0); return 0; } @@ -738,7 +750,7 @@ static void lguest_time_init(void) * will not tolerate us trying to use that), the stack pointer, and the number * of pages in the stack. */ static void lguest_load_sp0(struct tss_struct *tss, - struct thread_struct *thread) + struct thread_struct *thread) { lazy_hcall(LHCALL_SET_STACK, __KERNEL_DS|0x1, thread->sp0, THREAD_SIZE/PAGE_SIZE); @@ -786,9 +798,8 @@ static void lguest_safe_halt(void) hcall(LHCALL_HALT, 0, 0, 0); } -/* Perhaps CRASH isn't the best name for this hypercall, but we use it to get a - * message out when we're crashing as well as elegant termination like powering - * off. +/* The SHUTDOWN hypercall takes a string to describe what's happening, and + * an argument which says whether this to restart (reboot) the Guest or not. * * Note that the Host always prefers that the Guest speak in physical addresses * rather than virtual addresses, so we use __pa() here. */ @@ -816,8 +827,9 @@ static struct notifier_block paniced = { /* Setting up memory is fairly easy. */ static __init char *lguest_memory_setup(void) { - /* We do this here and not earlier because lockcheck barfs if we do it - * before start_kernel() */ + /* We do this here and not earlier because lockcheck used to barf if we + * did it before start_kernel(). I think we fixed that, so it'd be + * nice to move it back to lguest_init. Patch welcome... */ atomic_notifier_chain_register(&panic_notifier_list, &paniced); /* The Linux bootloader header contains an "e820" memory map: the @@ -850,12 +862,19 @@ static __init int early_put_chars(u32 vtermno, const char *buf, int count) return len; } +/* Rebooting also tells the Host we're finished, but the RESTART flag tells the + * Launcher to reboot us. */ +static void lguest_restart(char *reason) +{ + hcall(LHCALL_SHUTDOWN, __pa(reason), LGUEST_SHUTDOWN_RESTART, 0); +} + /*G:050 * Patching (Powerfully Placating Performance Pedants) * - * We have already seen that pv_ops structures let us replace simple - * native instructions with calls to the appropriate back end all throughout - * the kernel. This allows the same kernel to run as a Guest and as a native + * We have already seen that pv_ops structures let us replace simple native + * instructions with calls to the appropriate back end all throughout the + * kernel. This allows the same kernel to run as a Guest and as a native * kernel, but it's slow because of all the indirect branches. * * Remember that David Wheeler quote about "Any problem in computer science can @@ -908,14 +927,9 @@ static unsigned lguest_patch(u8 type, u16 clobber, void *ibuf, return insn_len; } -static void lguest_restart(char *reason) -{ - hcall(LHCALL_SHUTDOWN, __pa(reason), LGUEST_SHUTDOWN_RESTART, 0); -} - -/*G:030 Once we get to lguest_init(), we know we're a Guest. The pv_ops - * structures in the kernel provide points for (almost) every routine we have - * to override to avoid privileged instructions. */ +/*G:030 Once we get to lguest_init(), we know we're a Guest. The various + * pv_ops structures in the kernel provide points for (almost) every routine we + * have to override to avoid privileged instructions. */ __init void lguest_init(void) { /* We're under lguest, paravirt is enabled, and we're running at @@ -1003,9 +1017,9 @@ __init void lguest_init(void) * the normal data segment to get through booting. */ asm volatile ("mov %0, %%fs" : : "r" (__KERNEL_DS) : "memory"); - /* The Host uses the top of the Guest's virtual address space for the - * Host<->Guest Switcher, and it tells us how big that is in - * lguest_data.reserve_mem, set up on the LGUEST_INIT hypercall. */ + /* The Host<->Guest Switcher lives at the top of our address space, and + * the Host told us how big it is when we made LGUEST_INIT hypercall: + * it put the answer in lguest_data.reserve_mem */ reserve_top_address(lguest_data.reserve_mem); /* If we don't initialize the lock dependency checker now, it crashes @@ -1027,6 +1041,7 @@ __init void lguest_init(void) /* Math is always hard! */ new_cpu_data.hard_math = 1; + /* We don't have features. We have puppies! Puppies! */ #ifdef CONFIG_X86_MCE mce_disabled = 1; #endif @@ -1044,10 +1059,11 @@ __init void lguest_init(void) virtio_cons_early_init(early_put_chars); /* Last of all, we set the power management poweroff hook to point to - * the Guest routine to power off. */ + * the Guest routine to power off, and the reboot hook to our restart + * routine. */ pm_power_off = lguest_power_off; - machine_ops.restart = lguest_restart; + /* Now we're set up, call start_kernel() in init/main.c and we proceed * to boot as normal. It never returns. */ start_kernel(); diff --git a/arch/x86/lguest/i386_head.S b/arch/x86/lguest/i386_head.S index 95b6fbcded63..5c7cef34c9e7 100644 --- a/arch/x86/lguest/i386_head.S +++ b/arch/x86/lguest/i386_head.S @@ -5,13 +5,20 @@ #include #include -/*G:020 This is where we begin: head.S notes that the boot header's platform - * type field is "1" (lguest), so calls us here. +/*G:020 Our story starts with the kernel booting into startup_32 in + * arch/x86/kernel/head_32.S. It expects a boot header, which is created by + * the bootloader (the Launcher in our case). + * + * The startup_32 function does very little: it clears the uninitialized global + * C variables which we expect to be zero (ie. BSS) and then copies the boot + * header and kernel command line somewhere safe. Finally it checks the + * 'hardware_subarch' field. This was introduced in 2.6.24 for lguest and Xen: + * if it's set to '1' (lguest's assigned number), then it calls us here. * * WARNING: be very careful here! We're running at addresses equal to physical * addesses (around 0), not above PAGE_OFFSET as most code expectes * (eg. 0xC0000000). Jumps are relative, so they're OK, but we can't touch any - * data. + * data without remembering to subtract __PAGE_OFFSET! * * The .section line puts this code in .init.text so it will be discarded after * boot. */ @@ -24,7 +31,7 @@ ENTRY(lguest_entry) int $LGUEST_TRAP_ENTRY /* The Host put the toplevel pagetable in lguest_data.pgdir. The movsl - * instruction uses %esi implicitly as the source for the copy we' + * instruction uses %esi implicitly as the source for the copy we're * about to do. */ movl lguest_data - __PAGE_OFFSET + LGUEST_DATA_pgdir, %esi diff --git a/drivers/lguest/core.c b/drivers/lguest/core.c index c632c08cbbdc..5eea4356d703 100644 --- a/drivers/lguest/core.c +++ b/drivers/lguest/core.c @@ -1,8 +1,6 @@ /*P:400 This contains run_guest() which actually calls into the Host<->Guest * Switcher and analyzes the return, such as determining if the Guest wants the - * Host to do something. This file also contains useful helper routines, and a - * couple of non-obvious setup and teardown pieces which were implemented after - * days of debugging pain. :*/ + * Host to do something. This file also contains useful helper routines. :*/ #include #include #include @@ -49,8 +47,8 @@ static __init int map_switcher(void) * easy. */ - /* We allocate an array of "struct page"s. map_vm_area() wants the - * pages in this form, rather than just an array of pointers. */ + /* We allocate an array of struct page pointers. map_vm_area() wants + * this, rather than just an array of pages. */ switcher_page = kmalloc(sizeof(switcher_page[0])*TOTAL_SWITCHER_PAGES, GFP_KERNEL); if (!switcher_page) { @@ -172,7 +170,7 @@ void __lgread(struct lg_cpu *cpu, void *b, unsigned long addr, unsigned bytes) } } -/* This is the write (copy into guest) version. */ +/* This is the write (copy into Guest) version. */ void __lgwrite(struct lg_cpu *cpu, unsigned long addr, const void *b, unsigned bytes) { @@ -209,9 +207,9 @@ int run_guest(struct lg_cpu *cpu, unsigned long __user *user) if (cpu->break_out) return -EAGAIN; - /* Check if there are any interrupts which can be delivered - * now: if so, this sets up the hander to be executed when we - * next run the Guest. */ + /* Check if there are any interrupts which can be delivered now: + * if so, this sets up the hander to be executed when we next + * run the Guest. */ maybe_do_interrupt(cpu); /* All long-lived kernel loops need to check with this horrible @@ -246,8 +244,10 @@ int run_guest(struct lg_cpu *cpu, unsigned long __user *user) lguest_arch_handle_trap(cpu); } + /* Special case: Guest is 'dead' but wants a reboot. */ if (cpu->lg->dead == ERR_PTR(-ERESTART)) return -ERESTART; + /* The Guest is dead => "No such file or directory" */ return -ENOENT; } diff --git a/drivers/lguest/hypercalls.c b/drivers/lguest/hypercalls.c index 0f2cb4fd7c69..54d66f05fefa 100644 --- a/drivers/lguest/hypercalls.c +++ b/drivers/lguest/hypercalls.c @@ -29,7 +29,7 @@ #include "lg.h" /*H:120 This is the core hypercall routine: where the Guest gets what it wants. - * Or gets killed. Or, in the case of LHCALL_CRASH, both. */ + * Or gets killed. Or, in the case of LHCALL_SHUTDOWN, both. */ static void do_hcall(struct lg_cpu *cpu, struct hcall_args *args) { switch (args->arg0) { @@ -190,6 +190,13 @@ static void initialize(struct lg_cpu *cpu) * pagetable. */ guest_pagetable_clear_all(cpu); } +/*:*/ + +/*M:013 If a Guest reads from a page (so creates a mapping) that it has never + * written to, and then the Launcher writes to it (ie. the output of a virtual + * device), the Guest will still see the old page. In practice, this never + * happens: why would the Guest read a page which it has never written to? But + * a similar scenario might one day bite us, so it's worth mentioning. :*/ /*H:100 * Hypercalls @@ -227,7 +234,7 @@ void do_hypercalls(struct lg_cpu *cpu) * However, if we are signalled or the Guest sends I/O to the * Launcher, the run_guest() loop will exit without running the * Guest. When it comes back it would try to re-run the - * hypercall. */ + * hypercall. Finding that bug sucked. */ cpu->hcall = NULL; } } diff --git a/drivers/lguest/interrupts_and_traps.c b/drivers/lguest/interrupts_and_traps.c index 32e97c1858e5..0414ddf87587 100644 --- a/drivers/lguest/interrupts_and_traps.c +++ b/drivers/lguest/interrupts_and_traps.c @@ -144,7 +144,6 @@ void maybe_do_interrupt(struct lg_cpu *cpu) if (copy_from_user(&blk, cpu->lg->lguest_data->blocked_interrupts, sizeof(blk))) return; - bitmap_andnot(blk, cpu->irqs_pending, blk, LGUEST_IRQS); /* Find the first interrupt. */ @@ -237,9 +236,9 @@ void free_interrupts(void) clear_bit(syscall_vector, used_vectors); } -/*H:220 Now we've got the routines to deliver interrupts, delivering traps - * like page fault is easy. The only trick is that Intel decided that some - * traps should have error codes: */ +/*H:220 Now we've got the routines to deliver interrupts, delivering traps like + * page fault is easy. The only trick is that Intel decided that some traps + * should have error codes: */ static int has_err(unsigned int trap) { return (trap == 8 || (trap >= 10 && trap <= 14) || trap == 17); diff --git a/drivers/lguest/lguest_device.c b/drivers/lguest/lguest_device.c index 1b2ec0bf5eb1..2bc9bf7e88e5 100644 --- a/drivers/lguest/lguest_device.c +++ b/drivers/lguest/lguest_device.c @@ -1,10 +1,10 @@ /*P:050 Lguest guests use a very simple method to describe devices. It's a - * series of device descriptors contained just above the top of normal + * series of device descriptors contained just above the top of normal Guest * memory. * * We use the standard "virtio" device infrastructure, which provides us with a * console, a network and a block driver. Each one expects some configuration - * information and a "virtqueue" mechanism to send and receive data. :*/ + * information and a "virtqueue" or two to send and receive data. :*/ #include #include #include @@ -53,7 +53,7 @@ struct lguest_device { * Device configurations * * The configuration information for a device consists of one or more - * virtqueues, a feature bitmaks, and some configuration bytes. The + * virtqueues, a feature bitmap, and some configuration bytes. The * configuration bytes don't really matter to us: the Launcher sets them up, and * the driver will look at them during setup. * @@ -179,7 +179,7 @@ struct lguest_vq_info }; /* When the virtio_ring code wants to prod the Host, it calls us here and we - * make a hypercall. We hand the page number of the virtqueue so the Host + * make a hypercall. We hand the physical address of the virtqueue so the Host * knows which virtqueue we're talking about. */ static void lg_notify(struct virtqueue *vq) { @@ -199,7 +199,8 @@ static void lg_notify(struct virtqueue *vq) * allocate its own pages and tell the Host where they are, but for lguest it's * simpler for the Host to simply tell us where the pages are. * - * So we provide devices with a "find virtqueue and set it up" function. */ + * So we provide drivers with a "find the Nth virtqueue and set it up" + * function. */ static struct virtqueue *lg_find_vq(struct virtio_device *vdev, unsigned index, void (*callback)(struct virtqueue *vq)) diff --git a/drivers/lguest/lguest_user.c b/drivers/lguest/lguest_user.c index 2221485b0773..564e425d71dd 100644 --- a/drivers/lguest/lguest_user.c +++ b/drivers/lguest/lguest_user.c @@ -73,7 +73,7 @@ static ssize_t read(struct file *file, char __user *user, size_t size,loff_t*o) if (current != cpu->tsk) return -EPERM; - /* If the guest is already dead, we indicate why */ + /* If the Guest is already dead, we indicate why */ if (lg->dead) { size_t len; @@ -88,7 +88,7 @@ static ssize_t read(struct file *file, char __user *user, size_t size,loff_t*o) return len; } - /* If we returned from read() last time because the Guest notified, + /* If we returned from read() last time because the Guest sent I/O, * clear the flag. */ if (cpu->pending_notify) cpu->pending_notify = 0; @@ -97,14 +97,20 @@ static ssize_t read(struct file *file, char __user *user, size_t size,loff_t*o) return run_guest(cpu, (unsigned long __user *)user); } +/*L:025 This actually initializes a CPU. For the moment, a Guest is only + * uniprocessor, so "id" is always 0. */ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) { + /* We have a limited number the number of CPUs in the lguest struct. */ if (id >= NR_CPUS) return -EINVAL; + /* Set up this CPU's id, and pointer back to the lguest struct. */ cpu->id = id; cpu->lg = container_of((cpu - id), struct lguest, cpus[0]); cpu->lg->nr_cpus++; + + /* Each CPU has a timer it can set. */ init_clockdev(cpu); /* We need a complete page for the Guest registers: they are accessible @@ -120,11 +126,11 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) * address. */ lguest_arch_setup_regs(cpu, start_ip); - /* Initialize the queue for the waker to wait on */ + /* Initialize the queue for the Waker to wait on */ init_waitqueue_head(&cpu->break_wq); /* We keep a pointer to the Launcher task (ie. current task) for when - * other Guests want to wake this one (inter-Guest I/O). */ + * other Guests want to wake this one (eg. console input). */ cpu->tsk = current; /* We need to keep a pointer to the Launcher's memory map, because if @@ -136,6 +142,7 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) * when the same Guest runs on the same CPU twice. */ cpu->last_pages = NULL; + /* No error == success. */ return 0; } @@ -185,14 +192,13 @@ static int initialize(struct file *file, const unsigned long __user *input) lg->mem_base = (void __user *)(long)args[0]; lg->pfn_limit = args[1]; - /* This is the first cpu */ + /* This is the first cpu (cpu 0) and it will start booting at args[3] */ err = lg_cpu_start(&lg->cpus[0], 0, args[3]); if (err) goto release_guest; /* Initialize the Guest's shadow page tables, using the toplevel - * address the Launcher gave us. This allocates memory, so can - * fail. */ + * address the Launcher gave us. This allocates memory, so can fail. */ err = init_guest_pagetable(lg, args[2]); if (err) goto free_regs; @@ -218,11 +224,16 @@ unlock: /*L:010 The first operation the Launcher does must be a write. All writes * start with an unsigned long number: for the first write this must be * LHREQ_INITIALIZE to set up the Guest. After that the Launcher can use - * writes of other values to send interrupts. */ + * writes of other values to send interrupts. + * + * Note that we overload the "offset" in the /dev/lguest file to indicate what + * CPU number we're dealing with. Currently this is always 0, since we only + * support uniprocessor Guests, but you can see the beginnings of SMP support + * here. */ static ssize_t write(struct file *file, const char __user *in, size_t size, loff_t *off) { - /* Once the guest is initialized, we hold the "struct lguest" in the + /* Once the Guest is initialized, we hold the "struct lguest" in the * file private data. */ struct lguest *lg = file->private_data; const unsigned long __user *input = (const unsigned long __user *)in; @@ -230,6 +241,7 @@ static ssize_t write(struct file *file, const char __user *in, struct lg_cpu *uninitialized_var(cpu); unsigned int cpu_id = *off; + /* The first value tells us what this request is. */ if (get_user(req, input) != 0) return -EFAULT; input++; diff --git a/drivers/lguest/page_tables.c b/drivers/lguest/page_tables.c index a7f64a9d67e0..d93500f24fbb 100644 --- a/drivers/lguest/page_tables.c +++ b/drivers/lguest/page_tables.c @@ -2,8 +2,8 @@ * previous encounters. It's functional, and as neat as it can be in the * circumstances, but be wary, for these things are subtle and break easily. * The Guest provides a virtual to physical mapping, but we can neither trust - * it nor use it: we verify and convert it here to point the hardware to the - * actual Guest pages when running the Guest. :*/ + * it nor use it: we verify and convert it here then point the CPU to the + * converted Guest pages when running the Guest. :*/ /* Copyright (C) Rusty Russell IBM Corporation 2006. * GPL v2 and any later version */ @@ -106,6 +106,11 @@ static unsigned long gpte_addr(pgd_t gpgd, unsigned long vaddr) BUG_ON(!(pgd_flags(gpgd) & _PAGE_PRESENT)); return gpage + ((vaddr>>PAGE_SHIFT) % PTRS_PER_PTE) * sizeof(pte_t); } +/*:*/ + +/*M:014 get_pfn is slow; it takes the mmap sem and calls get_user_pages. We + * could probably try to grab batches of pages here as an optimization + * (ie. pre-faulting). :*/ /*H:350 This routine takes a page number given by the Guest and converts it to * an actual, physical page number. It can fail for several reasons: the @@ -113,8 +118,8 @@ static unsigned long gpte_addr(pgd_t gpgd, unsigned long vaddr) * and the page is read-only, or the write flag was set and the page was * shared so had to be copied, but we ran out of memory. * - * This holds a reference to the page, so release_pte() is careful to - * put that back. */ + * This holds a reference to the page, so release_pte() is careful to put that + * back. */ static unsigned long get_pfn(unsigned long virtpfn, int write) { struct page *page; @@ -532,13 +537,13 @@ static void do_set_pte(struct lg_cpu *cpu, int idx, * all processes. So when the page table above that address changes, we update * all the page tables, not just the current one. This is rare. * - * The benefit is that when we have to track a new page table, we can copy keep - * all the kernel mappings. This speeds up context switch immensely. */ + * The benefit is that when we have to track a new page table, we can keep all + * the kernel mappings. This speeds up context switch immensely. */ void guest_set_pte(struct lg_cpu *cpu, unsigned long gpgdir, unsigned long vaddr, pte_t gpte) { - /* Kernel mappings must be changed on all top levels. Slow, but - * doesn't happen often. */ + /* Kernel mappings must be changed on all top levels. Slow, but doesn't + * happen often. */ if (vaddr >= cpu->lg->kernel_address) { unsigned int i; for (i = 0; i < ARRAY_SIZE(cpu->lg->pgdirs); i++) @@ -704,12 +709,11 @@ static __init void populate_switcher_pte_page(unsigned int cpu, /* We've made it through the page table code. Perhaps our tired brains are * still processing the details, or perhaps we're simply glad it's over. * - * If nothing else, note that all this complexity in juggling shadow page - * tables in sync with the Guest's page tables is for one reason: for most - * Guests this page table dance determines how bad performance will be. This - * is why Xen uses exotic direct Guest pagetable manipulation, and why both - * Intel and AMD have implemented shadow page table support directly into - * hardware. + * If nothing else, note that all this complexity in juggling shadow page tables + * in sync with the Guest's page tables is for one reason: for most Guests this + * page table dance determines how bad performance will be. This is why Xen + * uses exotic direct Guest pagetable manipulation, and why both Intel and AMD + * have implemented shadow page table support directly into hardware. * * There is just one file remaining in the Host. */ diff --git a/drivers/lguest/x86/core.c b/drivers/lguest/x86/core.c index 635187812d52..5126d5d9ea0e 100644 --- a/drivers/lguest/x86/core.c +++ b/drivers/lguest/x86/core.c @@ -17,6 +17,13 @@ * along with this program; if not, write to the Free Software * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ +/*P:450 This file contains the x86-specific lguest code. It used to be all + * mixed in with drivers/lguest/core.c but several foolhardy code slashers + * wrestled most of the dependencies out to here in preparation for porting + * lguest to other architectures (see what I mean by foolhardy?). + * + * This also contains a couple of non-obvious setup and teardown pieces which + * were implemented after days of debugging pain. :*/ #include #include #include @@ -157,6 +164,8 @@ static void run_guest_once(struct lg_cpu *cpu, struct lguest_pages *pages) * also simplify copy_in_guest_info(). Note that we'd still need to restore * things when we exit to Launcher userspace, but that's fairly easy. * + * We could also try using this hooks for PGE, but that might be too expensive. + * * The hooks were designed for KVM, but we can also put them to good use. :*/ /*H:040 This is the i386-specific code to setup and run the Guest. Interrupts @@ -182,7 +191,7 @@ void lguest_arch_run_guest(struct lg_cpu *cpu) * was doing. */ run_guest_once(cpu, lguest_pages(raw_smp_processor_id())); - /* Note that the "regs" pointer contains two extra entries which are + /* Note that the "regs" structure contains two extra entries which are * not really registers: a trap number which says what interrupt or * trap made the switcher code come back, and an error code which some * traps set. */ @@ -293,11 +302,10 @@ void lguest_arch_handle_trap(struct lg_cpu *cpu) break; case 14: /* We've intercepted a Page Fault. */ /* The Guest accessed a virtual address that wasn't mapped. - * This happens a lot: we don't actually set up most of the - * page tables for the Guest at all when we start: as it runs - * it asks for more and more, and we set them up as - * required. In this case, we don't even tell the Guest that - * the fault happened. + * This happens a lot: we don't actually set up most of the page + * tables for the Guest at all when we start: as it runs it asks + * for more and more, and we set them up as required. In this + * case, we don't even tell the Guest that the fault happened. * * The errcode tells whether this was a read or a write, and * whether kernel or userspace code. */ @@ -342,7 +350,7 @@ void lguest_arch_handle_trap(struct lg_cpu *cpu) if (!deliver_trap(cpu, cpu->regs->trapnum)) /* If the Guest doesn't have a handler (either it hasn't * registered any yet, or it's one of the faults we don't let - * it handle), it dies with a cryptic error message. */ + * it handle), it dies with this cryptic error message. */ kill_guest(cpu, "unhandled trap %li at %#lx (%#lx)", cpu->regs->trapnum, cpu->regs->eip, cpu->regs->trapnum == 14 ? cpu->arch.last_pagefault @@ -375,8 +383,8 @@ void __init lguest_arch_host_init(void) * The only exception is the interrupt handlers in switcher.S: their * addresses are placed in a table (default_idt_entries), so we need to * update the table with the new addresses. switcher_offset() is a - * convenience function which returns the distance between the builtin - * switcher code and the high-mapped copy we just made. */ + * convenience function which returns the distance between the + * compiled-in switcher code and the high-mapped copy we just made. */ for (i = 0; i < IDT_ENTRIES; i++) default_idt_entries[i] += switcher_offset(); @@ -416,7 +424,7 @@ void __init lguest_arch_host_init(void) state->guest_gdt_desc.address = (long)&state->guest_gdt; /* We know where we want the stack to be when the Guest enters - * the switcher: in pages->regs. The stack grows upwards, so + * the Switcher: in pages->regs. The stack grows upwards, so * we start it at the end of that structure. */ state->guest_tss.sp0 = (long)(&pages->regs + 1); /* And this is the GDT entry to use for the stack: we keep a @@ -513,8 +521,8 @@ int lguest_arch_init_hypercalls(struct lg_cpu *cpu) { u32 tsc_speed; - /* The pointer to the Guest's "struct lguest_data" is the only - * argument. We check that address now. */ + /* The pointer to the Guest's "struct lguest_data" is the only argument. + * We check that address now. */ if (!lguest_address_ok(cpu->lg, cpu->hcall->arg1, sizeof(*cpu->lg->lguest_data))) return -EFAULT; @@ -546,6 +554,7 @@ int lguest_arch_init_hypercalls(struct lg_cpu *cpu) return 0; } +/*:*/ /*L:030 lguest_arch_setup_regs() * diff --git a/drivers/lguest/x86/switcher_32.S b/drivers/lguest/x86/switcher_32.S index 0af8baaa0d4a..3fc15318a80f 100644 --- a/drivers/lguest/x86/switcher_32.S +++ b/drivers/lguest/x86/switcher_32.S @@ -1,6 +1,6 @@ -/*P:900 This is the Switcher: code which sits at 0xFFC00000 to do the low-level - * Guest<->Host switch. It is as simple as it can be made, but it's naturally - * very specific to x86. +/*P:900 This is the Switcher: code which sits at 0xFFC00000 astride both the + * Host and Guest to do the low-level Guest<->Host switch. It is as simple as + * it can be made, but it's naturally very specific to x86. * * You have now completed Preparation. If this has whet your appetite; if you * are feeling invigorated and refreshed then the next, more challenging stage @@ -189,7 +189,7 @@ ENTRY(switch_to_guest) // Interrupts are turned back on: we are Guest. iret -// We treat two paths to switch back to the Host +// We tread two paths to switch back to the Host // Yet both must save Guest state and restore Host // So we put the routine in a macro. #define SWITCH_TO_HOST \ diff --git a/include/asm-x86/lguest_hcall.h b/include/asm-x86/lguest_hcall.h index 758b9a5d4539..f239e7069cab 100644 --- a/include/asm-x86/lguest_hcall.h +++ b/include/asm-x86/lguest_hcall.h @@ -27,7 +27,7 @@ #ifndef __ASSEMBLY__ #include -/*G:031 First, how does our Guest contact the Host to ask for privileged +/*G:031 But first, how does our Guest contact the Host to ask for privileged * operations? There are two ways: the direct way is to make a "hypercall", * to make requests of the Host Itself. * diff --git a/include/linux/lguest_launcher.h b/include/linux/lguest_launcher.h index 589be3e1f3ac..e7217dc58f39 100644 --- a/include/linux/lguest_launcher.h +++ b/include/linux/lguest_launcher.h @@ -16,6 +16,10 @@ * a new device, we simply need to write a new virtio driver and create support * for it in the Launcher: this code won't need to change. * + * Virtio devices are also used by kvm, so we can simply reuse their optimized + * device drivers. And one day when everyone uses virtio, my plan will be + * complete. Bwahahahah! + * * Devices are described by a simplified ID, a status byte, and some "config" * bytes which describe this device's configuration. This is placed by the * Launcher just above the top of physical memory: @@ -26,7 +30,7 @@ struct lguest_device_desc { /* The number of virtqueues (first in config array) */ __u8 num_vq; /* The number of bytes of feature bits. Multiply by 2: one for host - * features and one for guest acknowledgements. */ + * features and one for Guest acknowledgements. */ __u8 feature_len; /* The number of bytes of the config array after virtqueues. */ __u8 config_len; -- cgit v1.3 From 50fd4407b8bfbde7c1a0bfe4f24de7df37164342 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Thu, 27 Mar 2008 17:42:50 -0700 Subject: [NET]: Use local_irq_{save,restore}() in napi_complete(). Based upon a lockdep report. Since ->poll() can be invoked from netpoll with interrupts disabled, we must not unconditionally enable interrupts in napi_complete(). Instead we must use local_irq_{save,restore}(). Noticed by Peter Zijlstra: netpoll_poll() poll_napi() spin_trylock(&napi->poll_lock) poll_one_napi() napi->poll() := sky2_poll() napi_complete() local_irq_disable() local_irq_enable() <--- *BUG* irq_exit() do_softirq() net_rx_action() spin_lock(&napi->poll_lock) <--- Deadlock! Because we still hold the lock.... Signed-off-by: David S. Miller --- include/linux/netdevice.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index a2f003239c85..fae6a7ececdb 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -383,9 +383,11 @@ static inline void __napi_complete(struct napi_struct *n) static inline void napi_complete(struct napi_struct *n) { - local_irq_disable(); + unsigned long flags; + + local_irq_save(flags); __napi_complete(n); - local_irq_enable(); + local_irq_restore(flags); } /** -- cgit v1.3 From 7c4b93d8269b9d35971a8239426b1f6ddc3d5ef7 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Fri, 21 Mar 2008 23:59:49 -0400 Subject: [PATCH] count ghost references to vfsmounts make propagate_mount_busy() exclude references from the vfsmounts that had been isolated by umount_tree() and are just waiting for release_mounts() to dispose of their ->mnt_parent/->mnt_mountpoint. Signed-off-by: Al Viro --- fs/namespace.c | 5 ++++- fs/pnode.c | 2 +- include/linux/mount.h | 1 + 3 files changed, 6 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/fs/namespace.c b/fs/namespace.c index 6324dfc80dc6..c175218ebae1 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -548,6 +548,7 @@ void release_mounts(struct list_head *head) m = mnt->mnt_parent; mnt->mnt_mountpoint = mnt->mnt_root; mnt->mnt_parent = mnt; + m->mnt_ghosts--; spin_unlock(&vfsmount_lock); dput(dentry); mntput(m); @@ -572,8 +573,10 @@ void umount_tree(struct vfsmount *mnt, int propagate, struct list_head *kill) __touch_mnt_namespace(p->mnt_ns); p->mnt_ns = NULL; list_del_init(&p->mnt_child); - if (p->mnt_parent != p) + if (p->mnt_parent != p) { + p->mnt_parent->mnt_ghosts++; p->mnt_mountpoint->d_mounted--; + } change_mnt_propagation(p, MS_PRIVATE); } } diff --git a/fs/pnode.c b/fs/pnode.c index 05ba692bc540..1d8f5447f3f7 100644 --- a/fs/pnode.c +++ b/fs/pnode.c @@ -225,7 +225,7 @@ out: */ static inline int do_refcount_check(struct vfsmount *mnt, int count) { - int mycount = atomic_read(&mnt->mnt_count); + int mycount = atomic_read(&mnt->mnt_count) - mnt->mnt_ghosts; return (mycount > count); } diff --git a/include/linux/mount.h b/include/linux/mount.h index 6d3047d8c91c..dac5e67ff3ee 100644 --- a/include/linux/mount.h +++ b/include/linux/mount.h @@ -61,6 +61,7 @@ struct vfsmount { atomic_t mnt_count; int mnt_expiry_mark; /* true if marked for expiry */ int mnt_pinned; + int mnt_ghosts; }; static inline struct vfsmount *mntget(struct vfsmount *mnt) -- cgit v1.3 From c35038becad0adb0e25261fff66d85b1a6ddd0c2 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 22 Mar 2008 00:46:23 -0400 Subject: [PATCH] do shrink_submounts() for all fs types ... and take it out of ->umount_begin() instances. Call with all locks already taken (by do_umount()) and leave calling release_mounts() to caller (it will do release_mounts() anyway, so we can just put into the same list). Signed-off-by: Al Viro --- fs/afs/internal.h | 1 - fs/afs/mntpt.c | 8 -------- fs/afs/super.c | 1 - fs/cifs/cifs_dfs_ref.c | 1 - fs/namespace.c | 23 ++++++++++------------- fs/nfs/super.c | 2 -- include/linux/mount.h | 1 - 7 files changed, 10 insertions(+), 27 deletions(-) (limited to 'include/linux') diff --git a/fs/afs/internal.h b/fs/afs/internal.h index 5ca3625cd39e..9ba16edc0af2 100644 --- a/fs/afs/internal.h +++ b/fs/afs/internal.h @@ -573,7 +573,6 @@ extern const struct file_operations afs_mntpt_file_operations; extern int afs_mntpt_check_symlink(struct afs_vnode *, struct key *); extern void afs_mntpt_kill_timer(void); -extern void afs_umount_begin(struct vfsmount *, int); /* * proc.c diff --git a/fs/afs/mntpt.c b/fs/afs/mntpt.c index a3510b8ba3e7..2f5503902c37 100644 --- a/fs/afs/mntpt.c +++ b/fs/afs/mntpt.c @@ -283,11 +283,3 @@ void afs_mntpt_kill_timer(void) cancel_delayed_work(&afs_mntpt_expiry_timer); flush_scheduled_work(); } - -/* - * begin unmount by attempting to remove all automounted mountpoints we added - */ -void afs_umount_begin(struct vfsmount *vfsmnt, int flags) -{ - shrink_submounts(vfsmnt, &afs_vfsmounts); -} diff --git a/fs/afs/super.c b/fs/afs/super.c index 36bbce45f44b..4b572b801d8d 100644 --- a/fs/afs/super.c +++ b/fs/afs/super.c @@ -50,7 +50,6 @@ static const struct super_operations afs_super_ops = { .write_inode = afs_write_inode, .destroy_inode = afs_destroy_inode, .clear_inode = afs_clear_inode, - .umount_begin = afs_umount_begin, .put_super = afs_put_super, .show_options = generic_show_options, }; diff --git a/fs/cifs/cifs_dfs_ref.c b/fs/cifs/cifs_dfs_ref.c index a1a95b027136..56c924033b78 100644 --- a/fs/cifs/cifs_dfs_ref.c +++ b/fs/cifs/cifs_dfs_ref.c @@ -33,7 +33,6 @@ void dfs_shrink_umount_helper(struct vfsmount *vfsmnt) { mark_mounts_for_expiry(&cifs_dfs_automount_list); mark_mounts_for_expiry(&cifs_dfs_automount_list); - shrink_submounts(vfsmnt, &cifs_dfs_automount_list); } /** diff --git a/fs/namespace.c b/fs/namespace.c index 1c78917ec930..7bd74b25930c 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -581,6 +581,8 @@ void umount_tree(struct vfsmount *mnt, int propagate, struct list_head *kill) } } +static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts); + static int do_umount(struct vfsmount *mnt, int flags) { struct super_block *sb = mnt->mnt_sb; @@ -653,6 +655,9 @@ static int do_umount(struct vfsmount *mnt, int flags) spin_lock(&vfsmount_lock); event++; + if (!(flags & MNT_DETACH)) + shrink_submounts(mnt, &umount_list); + retval = -EBUSY; if (flags & MNT_DETACH || !propagate_mount_busy(mnt, 2)) { if (!list_empty(&mnt->mnt_list)) @@ -1302,30 +1307,22 @@ resume: * process a list of expirable mountpoints with the intent of discarding any * submounts of a specific parent mountpoint */ -void shrink_submounts(struct vfsmount *mountpoint, struct list_head *mounts) +static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts) { LIST_HEAD(graveyard); - LIST_HEAD(umounts); - struct vfsmount *mnt; + struct vfsmount *m; - down_write(&namespace_sem); - spin_lock(&vfsmount_lock); /* extract submounts of 'mountpoint' from the expiration list */ - while (select_submounts(mountpoint, &graveyard)) { + while (select_submounts(mnt, &graveyard)) { while (!list_empty(&graveyard)) { - mnt = list_first_entry(&graveyard, struct vfsmount, + m = list_first_entry(&graveyard, struct vfsmount, mnt_expire); touch_mnt_namespace(mnt->mnt_ns); - umount_tree(mnt, 1, &umounts); + umount_tree(mnt, 1, umounts); } } - spin_unlock(&vfsmount_lock); - up_write(&namespace_sem); - release_mounts(&umounts); } -EXPORT_SYMBOL_GPL(shrink_submounts); - /* * Some copy_from_user() implementations do not return the exact number of * bytes remaining to copy on a fault. But copy_mount_options() requires that. diff --git a/fs/nfs/super.c b/fs/nfs/super.c index dd4dfcd632ec..f9219024f31a 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -589,8 +589,6 @@ static void nfs_umount_begin(struct vfsmount *vfsmnt, int flags) struct nfs_server *server = NFS_SB(vfsmnt->mnt_sb); struct rpc_clnt *rpc; - shrink_submounts(vfsmnt, &nfs_automount_list); - if (!(flags & MNT_FORCE)) return; /* -EIO all pending I/O */ diff --git a/include/linux/mount.h b/include/linux/mount.h index dac5e67ff3ee..5ee2df217cdf 100644 --- a/include/linux/mount.h +++ b/include/linux/mount.h @@ -99,7 +99,6 @@ extern int do_add_mount(struct vfsmount *newmnt, struct nameidata *nd, int mnt_flags, struct list_head *fslist); extern void mark_mounts_for_expiry(struct list_head *mounts); -extern void shrink_submounts(struct vfsmount *mountpoint, struct list_head *mounts); extern spinlock_t vfsmount_lock; extern dev_t name_to_dev_t(char *name); -- cgit v1.3 From 8c703d35fa91911dd92a18c31a718853f483ad80 Mon Sep 17 00:00:00 2001 From: Jonathan Corbet Date: Fri, 28 Mar 2008 14:15:49 -0700 Subject: in_atomic(): document why it is unsuitable for general use Discourage people from inappropriately using in_atomic() Signed-off-by: Jonathan Corbet Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/hardirq.h | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'include/linux') diff --git a/include/linux/hardirq.h b/include/linux/hardirq.h index 49829988bfa0..897f723bd222 100644 --- a/include/linux/hardirq.h +++ b/include/linux/hardirq.h @@ -72,6 +72,13 @@ #define in_softirq() (softirq_count()) #define in_interrupt() (irq_count()) +/* + * Are we running in atomic context? WARNING: this macro cannot + * always detect atomic context; in particular, it cannot know about + * held spinlocks in non-preemptible kernels. Thus it should not be + * used in the general case to determine whether sleeping is possible. + * Do not use in_atomic() in driver code. + */ #define in_atomic() ((preempt_count() & ~PREEMPT_ACTIVE) != 0) #ifdef CONFIG_PREEMPT -- cgit v1.3 From 3afe3925987adc3fc052abe404e44520c2072fc8 Mon Sep 17 00:00:00 2001 From: Harvey Harrison Date: Fri, 28 Mar 2008 14:16:01 -0700 Subject: kernel: add bit rotation helpers for 16 and 8 bit Will replace open-coded variants elsewhere. Done in the same style as the 32-bit versions. Signed-off-by: Harvey Harrison Acked-by: Randy Dunlap Cc: Johannes Berg Cc: John W. Linville Cc: Joe Perches Cc: Jiri Benc Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/bitops.h | 40 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+) (limited to 'include/linux') diff --git a/include/linux/bitops.h b/include/linux/bitops.h index 69c1edb9fe54..40d54731de7e 100644 --- a/include/linux/bitops.h +++ b/include/linux/bitops.h @@ -65,6 +65,46 @@ static inline __u32 ror32(__u32 word, unsigned int shift) return (word >> shift) | (word << (32 - shift)); } +/** + * rol16 - rotate a 16-bit value left + * @word: value to rotate + * @shift: bits to roll + */ +static inline __u16 rol16(__u16 word, unsigned int shift) +{ + return (word << shift) | (word >> (16 - shift)); +} + +/** + * ror16 - rotate a 16-bit value right + * @word: value to rotate + * @shift: bits to roll + */ +static inline __u16 ror16(__u16 word, unsigned int shift) +{ + return (word >> shift) | (word << (16 - shift)); +} + +/** + * rol8 - rotate an 8-bit value left + * @word: value to rotate + * @shift: bits to roll + */ +static inline __u8 rol8(__u8 word, unsigned int shift) +{ + return (word << shift) | (word >> (8 - shift)); +} + +/** + * ror8 - rotate an 8-bit value right + * @word: value to rotate + * @shift: bits to roll + */ +static inline __u8 ror8(__u8 word, unsigned int shift) +{ + return (word >> shift) | (word << (8 - shift)); +} + static inline unsigned fls_long(unsigned long l) { if (sizeof(l) == 4) -- cgit v1.3 From 5ac7ec85bcc70ef605657fb2d1106d27ab3bd131 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Fri, 28 Mar 2008 14:16:03 -0700 Subject: ext3: don't export ext3_fs.h and jbd.h Neither of the headers actually compiles when included from userpsace nor should it be made available as userspace tools should be using the libraries or at least headers from e2fsprogs. Signed-off-by: Christoph Hellwig Acked-by: "Theodore Ts'o" Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/Kbuild | 2 -- 1 file changed, 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/Kbuild b/include/linux/Kbuild index 4108b38ebb16..4a446a19295e 100644 --- a/include/linux/Kbuild +++ b/include/linux/Kbuild @@ -195,7 +195,6 @@ unifdef-y += ethtool.h unifdef-y += eventpoll.h unifdef-y += signalfd.h unifdef-y += ext2_fs.h -unifdef-y += ext3_fs.h unifdef-y += fb.h unifdef-y += fcntl.h unifdef-y += filter.h @@ -248,7 +247,6 @@ unifdef-y += isdn.h unifdef-y += isdnif.h unifdef-y += isdn_divertif.h unifdef-y += isdn_ppp.h -unifdef-y += jbd.h unifdef-y += joystick.h unifdef-y += kdev_t.h unifdef-y += kd.h -- cgit v1.3 From 0ef4730927c54cac5006759fa0cf50fbeff9d191 Mon Sep 17 00:00:00 2001 From: Matti Linnanvuori Date: Fri, 28 Mar 2008 16:33:00 -0700 Subject: net: Comment dev_kfree_skb_irq and dev_kfree_skb_any better Comment dev_kfree_skb_irq and dev_kfree_skb_any better. Signed-off-by: Matti Linnanvuori Signed-off-by: David S. Miller --- include/linux/netdevice.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h index fae6a7ececdb..ee81906b5164 100644 --- a/include/linux/netdevice.h +++ b/include/linux/netdevice.h @@ -1074,12 +1074,14 @@ static inline int netif_is_multiqueue(const struct net_device *dev) } /* Use this variant when it is known for sure that it - * is executing from interrupt context. + * is executing from hardware interrupt context or with hardware interrupts + * disabled. */ extern void dev_kfree_skb_irq(struct sk_buff *skb); /* Use this variant in places where it could be invoked - * either from interrupt or non-interrupt context. + * from either hardware interrupt or other context, with hardware interrupts + * either disabled or enabled. */ extern void dev_kfree_skb_any(struct sk_buff *skb); -- cgit v1.3 From 3ec25ebd69dc120d0590e64caaf1477aa88c8a93 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 27 Mar 2008 18:37:14 +0900 Subject: libata: ATA_EHI_LPM should be ATA_EH_LPM EH actions are ATA_EH_* not ATA_EHI_*. Rename ATA_EHI_LPM to ATA_EH_LPM. Signed-off-by: Tejun Heo Cc: Kristen Carlson Accardi Signed-off-by: Jeff Garzik --- drivers/ata/libata-core.c | 2 +- drivers/ata/libata-eh.c | 2 +- include/linux/libata.h | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) (limited to 'include/linux') diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c index c4248b37ff64..48519887f94a 100644 --- a/drivers/ata/libata-core.c +++ b/drivers/ata/libata-core.c @@ -972,7 +972,7 @@ static void ata_dev_disable_pm(struct ata_device *dev) void ata_lpm_schedule(struct ata_port *ap, enum link_pm policy) { ap->pm_policy = policy; - ap->link.eh_info.action |= ATA_EHI_LPM; + ap->link.eh_info.action |= ATA_EH_LPM; ap->link.eh_info.flags |= ATA_EHI_NO_AUTOPSY; ata_port_schedule_eh(ap); } diff --git a/drivers/ata/libata-eh.c b/drivers/ata/libata-eh.c index 681252fd8143..a5830329eda4 100644 --- a/drivers/ata/libata-eh.c +++ b/drivers/ata/libata-eh.c @@ -2748,7 +2748,7 @@ int ata_eh_recover(struct ata_port *ap, ata_prereset_fn_t prereset, ehc->i.flags &= ~ATA_EHI_SETMODE; } - if (ehc->i.action & ATA_EHI_LPM) + if (ehc->i.action & ATA_EH_LPM) ata_link_for_each_dev(dev, link) ata_dev_enable_pm(dev, ap->pm_policy); diff --git a/include/linux/libata.h b/include/linux/libata.h index 269cdba09578..b064bfeb69ee 100644 --- a/include/linux/libata.h +++ b/include/linux/libata.h @@ -295,6 +295,7 @@ enum { ATA_EH_SOFTRESET = (1 << 1), ATA_EH_HARDRESET = (1 << 2), ATA_EH_ENABLE_LINK = (1 << 3), + ATA_EH_LPM = (1 << 4), /* link power management action */ ATA_EH_RESET_MASK = ATA_EH_SOFTRESET | ATA_EH_HARDRESET, ATA_EH_PERDEV_MASK = ATA_EH_REVALIDATE, @@ -304,7 +305,6 @@ enum { ATA_EHI_RESUME_LINK = (1 << 1), /* resume link (reset modifier) */ ATA_EHI_NO_AUTOPSY = (1 << 2), /* no autopsy */ ATA_EHI_QUIET = (1 << 3), /* be quiet */ - ATA_EHI_LPM = (1 << 4), /* link power management action */ ATA_EHI_DID_SOFTRESET = (1 << 16), /* already soft-reset this port */ ATA_EHI_DID_HARDRESET = (1 << 17), /* already soft-reset this port */ -- cgit v1.3 From 729d4de96a5c090e40a918a41f63b7fb1b27c240 Mon Sep 17 00:00:00 2001 From: Bartlomiej Zolnierkiewicz Date: Sat, 29 Mar 2008 19:55:17 +0100 Subject: ide: fix defining SUPPORT_VLB_SYNC We need to check for CONFIG_{CRIS,FRV} not {CRIS,FRV}. Signed-off-by: Bartlomiej Zolnierkiewicz --- include/linux/ide.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/ide.h b/include/linux/ide.h index a3b69c10d667..bc26b2f27359 100644 --- a/include/linux/ide.h +++ b/include/linux/ide.h @@ -26,7 +26,7 @@ #include #include -#if defined(CRIS) || defined(FRV) +#if defined(CONFIG_CRIS) || defined(CONFIG_FRV) # define SUPPORT_VLB_SYNC 0 #else # define SUPPORT_VLB_SYNC 1 -- cgit v1.3 From 7d61c4596d11d624efb4bbcbad01f9cf2b321162 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 29 Mar 2008 03:09:28 +0000 Subject: compat_sys_wait4() prototype misannotation Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- include/linux/compat.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/linux') diff --git a/include/linux/compat.h b/include/linux/compat.h index a671dbff7a1f..8fa7857e153b 100644 --- a/include/linux/compat.h +++ b/include/linux/compat.h @@ -192,8 +192,8 @@ asmlinkage long compat_sys_select(int n, compat_ulong_t __user *inp, struct compat_timeval __user *tvp); asmlinkage long compat_sys_wait4(compat_pid_t pid, - compat_uint_t *stat_addr, int options, - struct compat_rusage *ru); + compat_uint_t __user *stat_addr, int options, + struct compat_rusage __user *ru); #define BITS_PER_COMPAT_LONG (8*sizeof(compat_long_t)) -- cgit v1.3 From b2ddb9019ea13fb7b62d8e45adcc468376af0de7 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Sat, 29 Mar 2008 03:09:38 +0000 Subject: dma_page_list ->base_address is a userland pointer Signed-off-by: Al Viro Signed-off-by: Linus Torvalds --- include/linux/dmaengine.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h index 261e43a4c873..34d440698293 100644 --- a/include/linux/dmaengine.h +++ b/include/linux/dmaengine.h @@ -423,7 +423,7 @@ void dma_async_device_unregister(struct dma_device *device); /* --- Helper iov-locking functions --- */ struct dma_page_list { - char *base_address; + char __user *base_address; int nr_pages; struct page **pages; }; -- cgit v1.3 From a7097ff89c3204737a07eecbc83f9ae6002cc534 Mon Sep 17 00:00:00 2001 From: Dmitry Torokhov Date: Tue, 1 Apr 2008 00:22:53 -0400 Subject: Input: make sure input interfaces pin parent input devices Recent driver core change causes references to parent devices being dropped early, at device_del() time, as opposed to when all children are freed. This causes oops in evdev with grabbed devices. Take the reference to the parent input device ourselves to ensure that it stays around long enough. Signed-off-by: Dmitry Torokhov --- drivers/input/evdev.c | 6 ++---- drivers/input/joydev.c | 3 ++- drivers/input/mousedev.c | 3 ++- include/linux/input.h | 5 +++-- 4 files changed, 9 insertions(+), 8 deletions(-) (limited to 'include/linux') diff --git a/drivers/input/evdev.c b/drivers/input/evdev.c index 99562cee827e..b32984bc516f 100644 --- a/drivers/input/evdev.c +++ b/drivers/input/evdev.c @@ -124,6 +124,7 @@ static void evdev_free(struct device *dev) { struct evdev *evdev = container_of(dev, struct evdev, dev); + input_put_device(evdev->handle.dev); kfree(evdev); } @@ -853,9 +854,6 @@ static void evdev_cleanup(struct evdev *evdev) evdev_hangup(evdev); evdev_remove_chrdev(evdev); - if (evdev->grab) - evdev_ungrab(evdev, evdev->grab); - /* evdev is marked dead so no one else accesses evdev->open */ if (evdev->open) { input_flush_device(handle, NULL); @@ -896,7 +894,7 @@ static int evdev_connect(struct input_handler *handler, struct input_dev *dev, evdev->exist = 1; evdev->minor = minor; - evdev->handle.dev = dev; + evdev->handle.dev = input_get_device(dev); evdev->handle.name = evdev->name; evdev->handle.handler = handler; evdev->handle.private = evdev; diff --git a/drivers/input/joydev.c b/drivers/input/joydev.c index 22b2789ef58a..65d7077a75a1 100644 --- a/drivers/input/joydev.c +++ b/drivers/input/joydev.c @@ -171,6 +171,7 @@ static void joydev_free(struct device *dev) { struct joydev *joydev = container_of(dev, struct joydev, dev); + input_put_device(joydev->handle.dev); kfree(joydev); } @@ -750,7 +751,7 @@ static int joydev_connect(struct input_handler *handler, struct input_dev *dev, joydev->minor = minor; joydev->exist = 1; - joydev->handle.dev = dev; + joydev->handle.dev = input_get_device(dev); joydev->handle.name = joydev->name; joydev->handle.handler = handler; joydev->handle.private = joydev; diff --git a/drivers/input/mousedev.c b/drivers/input/mousedev.c index bbbe5e81adc1..b989748598ae 100644 --- a/drivers/input/mousedev.c +++ b/drivers/input/mousedev.c @@ -414,6 +414,7 @@ static void mousedev_free(struct device *dev) { struct mousedev *mousedev = container_of(dev, struct mousedev, dev); + input_put_device(mousedev->handle.dev); kfree(mousedev); } @@ -865,7 +866,7 @@ static struct mousedev *mousedev_create(struct input_dev *dev, mousedev->minor = minor; mousedev->exist = 1; - mousedev->handle.dev = dev; + mousedev->handle.dev = input_get_device(dev); mousedev->handle.name = mousedev->name; mousedev->handle.handler = handler; mousedev->handle.private = mousedev; diff --git a/include/linux/input.h b/include/linux/input.h index 1bdc39a8c76c..cae2c35d1206 100644 --- a/include/linux/input.h +++ b/include/linux/input.h @@ -1227,12 +1227,13 @@ void input_free_device(struct input_dev *dev); static inline struct input_dev *input_get_device(struct input_dev *dev) { - return to_input_dev(get_device(&dev->dev)); + return dev ? to_input_dev(get_device(&dev->dev)) : NULL; } static inline void input_put_device(struct input_dev *dev) { - put_device(&dev->dev); + if (dev) + put_device(&dev->dev); } static inline void *input_get_drvdata(struct input_dev *dev) -- cgit v1.3 From 34e6bbf23c8f43e8713d9bd092680f1660494b4a Mon Sep 17 00:00:00 2001 From: Fabio Checconi Date: Wed, 2 Apr 2008 14:31:02 +0200 Subject: cfq-iosched: fix rcu freeing of cfq io contexts SLAB_DESTROY_BY_RCU is not a direct substitute for normal call_rcu() freeing, since it'll page freeing but NOT object freeing. So change cfq to do the freeing on its own. Signed-off-by: Fabio Checconi Acked-by: Paul E. McKenney Signed-off-by: Jens Axboe --- block/cfq-iosched.c | 57 ++++++++++++++++++++++------------------------- include/linux/iocontext.h | 3 +++ 2 files changed, 30 insertions(+), 30 deletions(-) (limited to 'include/linux') diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c index 0f962ecae91f..f26da2bfcc15 100644 --- a/block/cfq-iosched.c +++ b/block/cfq-iosched.c @@ -1143,24 +1143,37 @@ static void cfq_put_queue(struct cfq_queue *cfqq) } /* - * Call func for each cic attached to this ioc. Returns number of cic's seen. + * Call func for each cic attached to this ioc. */ -static unsigned int +static void call_for_each_cic(struct io_context *ioc, void (*func)(struct io_context *, struct cfq_io_context *)) { struct cfq_io_context *cic; struct hlist_node *n; - int called = 0; rcu_read_lock(); - hlist_for_each_entry_rcu(cic, n, &ioc->cic_list, cic_list) { + hlist_for_each_entry_rcu(cic, n, &ioc->cic_list, cic_list) func(ioc, cic); - called++; - } rcu_read_unlock(); +} + +static void cfq_cic_free_rcu(struct rcu_head *head) +{ + struct cfq_io_context *cic; + + cic = container_of(head, struct cfq_io_context, rcu_head); + + kmem_cache_free(cfq_ioc_pool, cic); + elv_ioc_count_dec(ioc_count); + + if (ioc_gone && !elv_ioc_count_read(ioc_count)) + complete(ioc_gone); +} - return called; +static void cfq_cic_free(struct cfq_io_context *cic) +{ + call_rcu(&cic->rcu_head, cfq_cic_free_rcu); } static void cic_free_func(struct io_context *ioc, struct cfq_io_context *cic) @@ -1174,24 +1187,18 @@ static void cic_free_func(struct io_context *ioc, struct cfq_io_context *cic) hlist_del_rcu(&cic->cic_list); spin_unlock_irqrestore(&ioc->lock, flags); - kmem_cache_free(cfq_ioc_pool, cic); + cfq_cic_free(cic); } static void cfq_free_io_context(struct io_context *ioc) { - int freed; - /* - * ioc->refcount is zero here, so no more cic's are allowed to be - * linked into this ioc. So it should be ok to iterate over the known - * list, we will see all cic's since no new ones are added. + * ioc->refcount is zero here, or we are called from elv_unregister(), + * so no more cic's are allowed to be linked into this ioc. So it + * should be ok to iterate over the known list, we will see all cic's + * since no new ones are added. */ - freed = call_for_each_cic(ioc, cic_free_func); - - elv_ioc_count_mod(ioc_count, -freed); - - if (ioc_gone && !elv_ioc_count_read(ioc_count)) - complete(ioc_gone); + call_for_each_cic(ioc, cic_free_func); } static void cfq_exit_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq) @@ -1458,15 +1465,6 @@ cfq_get_queue(struct cfq_data *cfqd, int is_sync, struct io_context *ioc, return cfqq; } -static void cfq_cic_free(struct cfq_io_context *cic) -{ - kmem_cache_free(cfq_ioc_pool, cic); - elv_ioc_count_dec(ioc_count); - - if (ioc_gone && !elv_ioc_count_read(ioc_count)) - complete(ioc_gone); -} - /* * We drop cfq io contexts lazily, so we may find a dead one. */ @@ -2138,7 +2136,7 @@ static int __init cfq_slab_setup(void) if (!cfq_pool) goto fail; - cfq_ioc_pool = KMEM_CACHE(cfq_io_context, SLAB_DESTROY_BY_RCU); + cfq_ioc_pool = KMEM_CACHE(cfq_io_context, 0); if (!cfq_ioc_pool) goto fail; @@ -2286,7 +2284,6 @@ static void __exit cfq_exit(void) smp_wmb(); if (elv_ioc_count_read(ioc_count)) wait_for_completion(ioc_gone); - synchronize_rcu(); cfq_slab_kill(); } diff --git a/include/linux/iocontext.h b/include/linux/iocontext.h index 1b4ccf25b4d2..cac4b364cd40 100644 --- a/include/linux/iocontext.h +++ b/include/linux/iocontext.h @@ -2,6 +2,7 @@ #define IOCONTEXT_H #include +#include /* * This is the per-process anticipatory I/O scheduler state. @@ -54,6 +55,8 @@ struct cfq_io_context { void (*dtor)(struct io_context *); /* destructor */ void (*exit)(struct io_context *); /* called on task exit */ + + struct rcu_head rcu_head; }; /* -- cgit v1.3 From dd135ebbd2a6b5e07dadb66c4dd033bb69531051 Mon Sep 17 00:00:00 2001 From: Christian Borntraeger Date: Wed, 2 Apr 2008 13:04:40 -0700 Subject: kvm: provide kvm.h for all architecture: fixes headers_install MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Currently include/linux/kvm.h is not considered by make headers_install, because Kbuild cannot handle " unifdef-$(CONFIG_FOO) += foo.h. This problem was introduced by commit fb56dbb31c4738a3918db81fd24da732ce3b4ae6 Author: Avi Kivity Date: Sun Dec 2 10:50:06 2007 +0200 KVM: Export include/linux/kvm.h only if $ARCH actually supports KVM Currently, make headers_check barfs due to , which includes, not existing. Rather than add a zillion s, export kvm. only if the arch actually supports it. Signed-off-by: Avi Kivity which makes this an 2.6.25 regression. One way of solving the issue is to enhance Kbuild, but Avi and David conviced me, that changing headers_install is not the way to go. This patch changes the definition for linux/kvm.h to unifdef-y. If  unifdef-y is used for linux/kvm.h "make headers_check" will fail on all architectures without asm/kvm.h. Therefore, this patch also provides asm/kvm.h on all architectures. Signed-off-by: Christian Borntraeger Acked-by: Avi Kivity Cc: Sam Ravnborg Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/asm-alpha/kvm.h | 6 ++++++ include/asm-arm/kvm.h | 6 ++++++ include/asm-avr32/kvm.h | 6 ++++++ include/asm-blackfin/kvm.h | 6 ++++++ include/asm-cris/kvm.h | 6 ++++++ include/asm-frv/kvm.h | 6 ++++++ include/asm-generic/Kbuild.asm | 2 ++ include/asm-h8300/kvm.h | 6 ++++++ include/asm-ia64/kvm.h | 6 ++++++ include/asm-m32r/kvm.h | 6 ++++++ include/asm-m68k/kvm.h | 6 ++++++ include/asm-m68knommu/kvm.h | 6 ++++++ include/asm-mips/kvm.h | 6 ++++++ include/asm-mn10300/kvm.h | 6 ++++++ include/asm-parisc/kvm.h | 6 ++++++ include/asm-powerpc/kvm.h | 6 ++++++ include/asm-s390/kvm.h | 6 ++++++ include/asm-sh/kvm.h | 6 ++++++ include/asm-sparc/kvm.h | 6 ++++++ include/asm-sparc64/kvm.h | 6 ++++++ include/asm-um/kvm.h | 6 ++++++ include/asm-v850/kvm.h | 6 ++++++ include/asm-xtensa/kvm.h | 6 ++++++ include/linux/Kbuild | 2 +- 24 files changed, 135 insertions(+), 1 deletion(-) create mode 100644 include/asm-alpha/kvm.h create mode 100644 include/asm-arm/kvm.h create mode 100644 include/asm-avr32/kvm.h create mode 100644 include/asm-blackfin/kvm.h create mode 100644 include/asm-cris/kvm.h create mode 100644 include/asm-frv/kvm.h create mode 100644 include/asm-h8300/kvm.h create mode 100644 include/asm-ia64/kvm.h create mode 100644 include/asm-m32r/kvm.h create mode 100644 include/asm-m68k/kvm.h create mode 100644 include/asm-m68knommu/kvm.h create mode 100644 include/asm-mips/kvm.h create mode 100644 include/asm-mn10300/kvm.h create mode 100644 include/asm-parisc/kvm.h create mode 100644 include/asm-powerpc/kvm.h create mode 100644 include/asm-s390/kvm.h create mode 100644 include/asm-sh/kvm.h create mode 100644 include/asm-sparc/kvm.h create mode 100644 include/asm-sparc64/kvm.h create mode 100644 include/asm-um/kvm.h create mode 100644 include/asm-v850/kvm.h create mode 100644 include/asm-xtensa/kvm.h (limited to 'include/linux') diff --git a/include/asm-alpha/kvm.h b/include/asm-alpha/kvm.h new file mode 100644 index 000000000000..b9daec429689 --- /dev/null +++ b/include/asm-alpha/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_ALPHA_H +#define __LINUX_KVM_ALPHA_H + +/* alpha does not support KVM */ + +#endif diff --git a/include/asm-arm/kvm.h b/include/asm-arm/kvm.h new file mode 100644 index 000000000000..cb3c08cbcb9e --- /dev/null +++ b/include/asm-arm/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_ARM_H +#define __LINUX_KVM_ARM_H + +/* arm does not support KVM */ + +#endif diff --git a/include/asm-avr32/kvm.h b/include/asm-avr32/kvm.h new file mode 100644 index 000000000000..8c5777020e2c --- /dev/null +++ b/include/asm-avr32/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_AVR32_H +#define __LINUX_KVM_AVR32_H + +/* avr32 does not support KVM */ + +#endif diff --git a/include/asm-blackfin/kvm.h b/include/asm-blackfin/kvm.h new file mode 100644 index 000000000000..e3477d77c014 --- /dev/null +++ b/include/asm-blackfin/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_BLACKFIN_H +#define __LINUX_KVM_BLACKFIN_H + +/* blackfin does not support KVM */ + +#endif diff --git a/include/asm-cris/kvm.h b/include/asm-cris/kvm.h new file mode 100644 index 000000000000..c860f51149f0 --- /dev/null +++ b/include/asm-cris/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_CRIS_H +#define __LINUX_KVM_CRIS_H + +/* cris does not support KVM */ + +#endif diff --git a/include/asm-frv/kvm.h b/include/asm-frv/kvm.h new file mode 100644 index 000000000000..9c8a4f08d0a9 --- /dev/null +++ b/include/asm-frv/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_FRV_H +#define __LINUX_KVM_FRV_H + +/* frv does not support KVM */ + +#endif diff --git a/include/asm-generic/Kbuild.asm b/include/asm-generic/Kbuild.asm index fd9dcfd91c39..92a6d91d0c1a 100644 --- a/include/asm-generic/Kbuild.asm +++ b/include/asm-generic/Kbuild.asm @@ -1,3 +1,5 @@ +header-y += kvm.h + ifeq ($(wildcard include/asm-$(SRCARCH)/a.out.h),include/asm-$(SRCARCH)/a.out.h) unifdef-y += a.out.h endif diff --git a/include/asm-h8300/kvm.h b/include/asm-h8300/kvm.h new file mode 100644 index 000000000000..bdbed7b987e1 --- /dev/null +++ b/include/asm-h8300/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_H8300_H +#define __LINUX_KVM_H8300_H + +/* h8300 does not support KVM */ + +#endif diff --git a/include/asm-ia64/kvm.h b/include/asm-ia64/kvm.h new file mode 100644 index 000000000000..030d29b4b26b --- /dev/null +++ b/include/asm-ia64/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_IA64_H +#define __LINUX_KVM_IA64_H + +/* ia64 does not support KVM */ + +#endif diff --git a/include/asm-m32r/kvm.h b/include/asm-m32r/kvm.h new file mode 100644 index 000000000000..99a40515b77e --- /dev/null +++ b/include/asm-m32r/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_M32R_H +#define __LINUX_KVM_M32R_H + +/* m32r does not support KVM */ + +#endif diff --git a/include/asm-m68k/kvm.h b/include/asm-m68k/kvm.h new file mode 100644 index 000000000000..7ed27fce5240 --- /dev/null +++ b/include/asm-m68k/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_M68K_H +#define __LINUX_KVM_M68K_H + +/* m68k does not support KVM */ + +#endif diff --git a/include/asm-m68knommu/kvm.h b/include/asm-m68knommu/kvm.h new file mode 100644 index 000000000000..b49d4258dabb --- /dev/null +++ b/include/asm-m68knommu/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_M68KNOMMU_H +#define __LINUX_KVM_M68KNOMMU_H + +/* m68knommu does not support KVM */ + +#endif diff --git a/include/asm-mips/kvm.h b/include/asm-mips/kvm.h new file mode 100644 index 000000000000..093a5b7f796b --- /dev/null +++ b/include/asm-mips/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_MIPS_H +#define __LINUX_KVM_MIPS_H + +/* mips does not support KVM */ + +#endif diff --git a/include/asm-mn10300/kvm.h b/include/asm-mn10300/kvm.h new file mode 100644 index 000000000000..f6b609ff4a57 --- /dev/null +++ b/include/asm-mn10300/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_MN10300_H +#define __LINUX_KVM_MN10300_H + +/* mn10300 does not support KVM */ + +#endif diff --git a/include/asm-parisc/kvm.h b/include/asm-parisc/kvm.h new file mode 100644 index 000000000000..00cc45812547 --- /dev/null +++ b/include/asm-parisc/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_PARISC_H +#define __LINUX_KVM_PARISC_H + +/* parisc does not support KVM */ + +#endif diff --git a/include/asm-powerpc/kvm.h b/include/asm-powerpc/kvm.h new file mode 100644 index 000000000000..d1b530fbf8dd --- /dev/null +++ b/include/asm-powerpc/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_POWERPC_H +#define __LINUX_KVM_POWERPC_H + +/* powerpc does not support KVM */ + +#endif diff --git a/include/asm-s390/kvm.h b/include/asm-s390/kvm.h new file mode 100644 index 000000000000..573f2a351386 --- /dev/null +++ b/include/asm-s390/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_S390_H +#define __LINUX_KVM_S390_H + +/* s390 does not support KVM */ + +#endif diff --git a/include/asm-sh/kvm.h b/include/asm-sh/kvm.h new file mode 100644 index 000000000000..6af51dbab2d0 --- /dev/null +++ b/include/asm-sh/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_SH_H +#define __LINUX_KVM_SH_H + +/* sh does not support KVM */ + +#endif diff --git a/include/asm-sparc/kvm.h b/include/asm-sparc/kvm.h new file mode 100644 index 000000000000..2e5478da3819 --- /dev/null +++ b/include/asm-sparc/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_SPARC_H +#define __LINUX_KVM_SPARC_H + +/* sparc does not support KVM */ + +#endif diff --git a/include/asm-sparc64/kvm.h b/include/asm-sparc64/kvm.h new file mode 100644 index 000000000000..380537a77bf9 --- /dev/null +++ b/include/asm-sparc64/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_SPARC64_H +#define __LINUX_KVM_SPARC64_H + +/* sparc64 does not support KVM */ + +#endif diff --git a/include/asm-um/kvm.h b/include/asm-um/kvm.h new file mode 100644 index 000000000000..66aa77094551 --- /dev/null +++ b/include/asm-um/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_UM_H +#define __LINUX_KVM_UM_H + +/* um does not support KVM */ + +#endif diff --git a/include/asm-v850/kvm.h b/include/asm-v850/kvm.h new file mode 100644 index 000000000000..3f729b79febc --- /dev/null +++ b/include/asm-v850/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_V850_H +#define __LINUX_KVM_V850_H + +/* v850 does not support KVM */ + +#endif diff --git a/include/asm-xtensa/kvm.h b/include/asm-xtensa/kvm.h new file mode 100644 index 000000000000..bda4e331e98c --- /dev/null +++ b/include/asm-xtensa/kvm.h @@ -0,0 +1,6 @@ +#ifndef __LINUX_KVM_XTENSA_H +#define __LINUX_KVM_XTENSA_H + +/* xtensa does not support KVM */ + +#endif diff --git a/include/linux/Kbuild b/include/linux/Kbuild index 4a446a19295e..9cdd12a9e843 100644 --- a/include/linux/Kbuild +++ b/include/linux/Kbuild @@ -253,7 +253,7 @@ unifdef-y += kd.h unifdef-y += kernelcapi.h unifdef-y += kernel.h unifdef-y += keyboard.h -unifdef-$(CONFIG_HAVE_KVM) += kvm.h +unifdef-y += kvm.h unifdef-y += llc.h unifdef-y += loop.h unifdef-y += lp.h -- cgit v1.3 From 436d34b36202ef724778ded1e9cb10f8c37b32bc Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 2 Apr 2008 17:28:46 +0900 Subject: libata: uninline atapi_cmd_type() Uninline atapi_cmd_type(). It doesn't really have to be inline and more case will be added which need to access unexported libata variable. Signed-off-by: Tejun Heo Signed-off-by: Jeff Garzik --- drivers/ata/libata-core.c | 34 ++++++++++++++++++++++++++++++++++ include/linux/libata.h | 22 +--------------------- 2 files changed, 35 insertions(+), 21 deletions(-) (limited to 'include/linux') diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c index 2db5c9c9ca10..e9b69ba489d2 100644 --- a/drivers/ata/libata-core.c +++ b/drivers/ata/libata-core.c @@ -325,6 +325,39 @@ static void ata_force_horkage(struct ata_device *dev) } } +/** + * atapi_cmd_type - Determine ATAPI command type from SCSI opcode + * @opcode: SCSI opcode + * + * Determine ATAPI command type from @opcode. + * + * LOCKING: + * None. + * + * RETURNS: + * ATAPI_{READ|WRITE|READ_CD|PASS_THRU|MISC} + */ +int atapi_cmd_type(u8 opcode) +{ + switch (opcode) { + case GPCMD_READ_10: + case GPCMD_READ_12: + return ATAPI_READ; + + case GPCMD_WRITE_10: + case GPCMD_WRITE_12: + case GPCMD_WRITE_AND_VERIFY_10: + return ATAPI_WRITE; + + case GPCMD_READ_CD: + case GPCMD_READ_CD_MSF: + return ATAPI_READ_CD; + + default: + return ATAPI_MISC; + } +} + /** * ata_tf_to_fis - Convert ATA taskfile to SATA FIS structure * @tf: Taskfile to convert @@ -7774,6 +7807,7 @@ EXPORT_SYMBOL_GPL(ata_tf_read); EXPORT_SYMBOL_GPL(ata_noop_dev_select); EXPORT_SYMBOL_GPL(ata_std_dev_select); EXPORT_SYMBOL_GPL(sata_print_link_status); +EXPORT_SYMBOL_GPL(atapi_cmd_type); EXPORT_SYMBOL_GPL(ata_tf_to_fis); EXPORT_SYMBOL_GPL(ata_tf_from_fis); EXPORT_SYMBOL_GPL(ata_pack_xfermask); diff --git a/include/linux/libata.h b/include/linux/libata.h index b064bfeb69ee..92c64909ed25 100644 --- a/include/linux/libata.h +++ b/include/linux/libata.h @@ -849,6 +849,7 @@ extern unsigned int ata_dev_try_classify(struct ata_device *dev, int present, */ extern void ata_tf_load(struct ata_port *ap, const struct ata_taskfile *tf); extern void ata_tf_read(struct ata_port *ap, struct ata_taskfile *tf); +extern int atapi_cmd_type(u8 opcode); extern void ata_tf_to_fis(const struct ata_taskfile *tf, u8 pmp, int is_cmd, u8 *fis); extern void ata_tf_from_fis(const u8 *fis, struct ata_taskfile *tf); @@ -1379,27 +1380,6 @@ static inline int ata_try_flush_cache(const struct ata_device *dev) ata_id_has_flush_ext(dev->id); } -static inline int atapi_cmd_type(u8 opcode) -{ - switch (opcode) { - case GPCMD_READ_10: - case GPCMD_READ_12: - return ATAPI_READ; - - case GPCMD_WRITE_10: - case GPCMD_WRITE_12: - case GPCMD_WRITE_AND_VERIFY_10: - return ATAPI_WRITE; - - case GPCMD_READ_CD: - case GPCMD_READ_CD_MSF: - return ATAPI_READ_CD; - - default: - return ATAPI_MISC; - } -} - static inline unsigned int ac_err_mask(u8 status) { if (status & (ATA_BUSY | ATA_DRQ)) -- cgit v1.3 From e52dcc4899cf1b7601379c31542bd91cd2997a64 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Wed, 2 Apr 2008 17:35:19 +0900 Subject: libata: ATA_12/16 doesn't fall into ATAPI_MISC SAT passthrus don't really fit into ATAPI_MISC class. SAT passthru commands always transfer multiple of 512 bytes and variable length response is not allowed. This patch creates a separate category - ATAPI_PASS_THRU - for these. This fixes HSM violation on "hdparm -I". Signed-off-by: Tejun Heo Signed-off-by: Jeff Garzik --- drivers/ata/libata-core.c | 5 +++++ include/linux/libata.h | 3 ++- 2 files changed, 7 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c index e9b69ba489d2..be95fdb69726 100644 --- a/drivers/ata/libata-core.c +++ b/drivers/ata/libata-core.c @@ -353,6 +353,11 @@ int atapi_cmd_type(u8 opcode) case GPCMD_READ_CD_MSF: return ATAPI_READ_CD; + case ATA_16: + case ATA_12: + if (atapi_passthru16) + return ATAPI_PASS_THRU; + /* fall thru */ default: return ATAPI_MISC; } diff --git a/include/linux/libata.h b/include/linux/libata.h index 92c64909ed25..37ee881c42ac 100644 --- a/include/linux/libata.h +++ b/include/linux/libata.h @@ -350,7 +350,8 @@ enum { ATAPI_READ = 0, /* READs */ ATAPI_WRITE = 1, /* WRITEs */ ATAPI_READ_CD = 2, /* READ CD [MSF] */ - ATAPI_MISC = 3, /* the rest */ + ATAPI_PASS_THRU = 3, /* SAT pass-thru */ + ATAPI_MISC = 4, /* the rest */ }; enum ata_xfer_mask { -- cgit v1.3 From 5761d64b277c287a7520b868c32d656ef03374b4 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 4 Apr 2008 16:26:10 +0200 Subject: x86: revert assign IRQs to hpet timer The commits: commit 37a47db8d7f0f38dac5acf5a13abbc8f401707fa Author: Balaji Rao Date: Wed Jan 30 13:30:03 2008 +0100 x86: assign IRQs to HPET timers, fix and commit e3f37a54f690d3e64995ea7ecea08c5ab3070faf Author: Balaji Rao Date: Wed Jan 30 13:30:03 2008 +0100 x86: assign IRQs to HPET timers have been identified to cause a regression on some platforms due to the assignement of legacy IRQs which makes the legacy devices connected to those IRQs disfunctional. Revert them. This fixes http://bugzilla.kernel.org/show_bug.cgi?id=10382 Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 9 ++++++--- drivers/char/hpet.c | 51 +++++++------------------------------------------- include/linux/hpet.h | 2 +- 3 files changed, 14 insertions(+), 48 deletions(-) (limited to 'include/linux') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 235fd6c77504..36652ea1a265 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -133,13 +133,16 @@ static void hpet_reserve_platform_timers(unsigned long id) #ifdef CONFIG_HPET_EMULATE_RTC hpet_reserve_timer(&hd, 1); #endif + hd.hd_irq[0] = HPET_LEGACY_8254; hd.hd_irq[1] = HPET_LEGACY_RTC; - for (i = 2; i < nrtimers; timer++, i++) - hd.hd_irq[i] = (timer->hpet_config & Tn_INT_ROUTE_CNF_MASK) >> - Tn_INT_ROUTE_CNF_SHIFT; + for (i = 2; i < nrtimers; timer++, i++) + hd.hd_irq[i] = (timer->hpet_config & Tn_INT_ROUTE_CNF_MASK) >> + Tn_INT_ROUTE_CNF_SHIFT; + hpet_alloc(&hd); + } #else static void hpet_reserve_platform_timers(unsigned long id) { } diff --git a/drivers/char/hpet.c b/drivers/char/hpet.c index 465ad35ed38f..1399971be689 100644 --- a/drivers/char/hpet.c +++ b/drivers/char/hpet.c @@ -731,14 +731,14 @@ static unsigned long hpet_calibrate(struct hpets *hpetp) int hpet_alloc(struct hpet_data *hdp) { - u64 cap, mcfg, hpet_config; + u64 cap, mcfg; struct hpet_dev *devp; - u32 i, ntimer, irq; + u32 i, ntimer; struct hpets *hpetp; size_t siz; struct hpet __iomem *hpet; static struct hpets *last = NULL; - unsigned long period, irq_bitmap; + unsigned long period; unsigned long long temp; /* @@ -765,47 +765,11 @@ int hpet_alloc(struct hpet_data *hdp) hpetp->hp_hpet_phys = hdp->hd_phys_address; hpetp->hp_ntimer = hdp->hd_nirqs; - hpet = hpetp->hp_hpet; - - /* Assign IRQs statically for legacy devices */ - hpetp->hp_dev[0].hd_hdwirq = hdp->hd_irq[0]; - hpetp->hp_dev[1].hd_hdwirq = hdp->hd_irq[1]; - - /* Assign IRQs dynamically for the others */ - for (i = 2, devp = &hpetp->hp_dev[2]; i < hdp->hd_nirqs; i++, devp++) { - struct hpet_timer __iomem *timer; - timer = &hpet->hpet_timers[devp - hpetp->hp_dev]; + for (i = 0; i < hdp->hd_nirqs; i++) + hpetp->hp_dev[i].hd_hdwirq = hdp->hd_irq[i]; - /* Check if there's already an IRQ assigned to the timer */ - if (hdp->hd_irq[i]) { - hpetp->hp_dev[i].hd_hdwirq = hdp->hd_irq[i]; - continue; - } - - hpet_config = readq(&timer->hpet_config); - irq_bitmap = (hpet_config & Tn_INT_ROUTE_CAP_MASK) - >> Tn_INT_ROUTE_CAP_SHIFT; - if (!irq_bitmap) - irq = 0; /* No valid IRQ Assignable */ - else { - irq = find_first_bit(&irq_bitmap, 32); - do { - hpet_config |= irq << Tn_INT_ROUTE_CNF_SHIFT; - writeq(hpet_config, &timer->hpet_config); - - /* - * Verify whether we have written a valid - * IRQ number by reading it back again - */ - hpet_config = readq(&timer->hpet_config); - if (irq == (hpet_config & Tn_INT_ROUTE_CNF_MASK) - >> Tn_INT_ROUTE_CNF_SHIFT) - break; /* Success */ - } while ((irq = (find_next_bit(&irq_bitmap, 32, irq)))); - } - hpetp->hp_dev[i].hd_hdwirq = irq; - } + hpet = hpetp->hp_hpet; cap = readq(&hpet->hpet_cap); @@ -836,8 +800,7 @@ int hpet_alloc(struct hpet_data *hdp) hpetp->hp_which, hdp->hd_phys_address, hpetp->hp_ntimer > 1 ? "s" : ""); for (i = 0; i < hpetp->hp_ntimer; i++) - printk("%s %d", i > 0 ? "," : "", - hpetp->hp_dev[i].hd_hdwirq); + printk("%s %d", i > 0 ? "," : "", hdp->hd_irq[i]); printk("\n"); printk(KERN_INFO "hpet%u: %u %d-bit timers, %Lu Hz\n", diff --git a/include/linux/hpet.h b/include/linux/hpet.h index 9cd94bfd07e5..2dc29ce6c8e4 100644 --- a/include/linux/hpet.h +++ b/include/linux/hpet.h @@ -64,7 +64,7 @@ struct hpet { */ #define Tn_INT_ROUTE_CAP_MASK (0xffffffff00000000ULL) -#define Tn_INT_ROUTE_CAP_SHIFT (32UL) +#define Tn_INI_ROUTE_CAP_SHIFT (32UL) #define Tn_FSB_INT_DELCAP_MASK (0x8000UL) #define Tn_FSB_INT_DELCAP_SHIFT (15) #define Tn_FSB_EN_CNF_MASK (0x4000UL) -- cgit v1.3 From 8bab8dded67d026c39367bbd5e27d2f6c556c38e Mon Sep 17 00:00:00 2001 From: Paul Menage Date: Fri, 4 Apr 2008 14:29:57 -0700 Subject: cgroups: add cgroup support for enabling controllers at boot time The effects of cgroup_disable=foo are: - foo isn't auto-mounted if you mount all cgroups in a single hierarchy - foo isn't visible as an individually mountable subsystem As a result there will only ever be one call to foo->create(), at init time; all processes will stay in this group, and the group will never be mounted on a visible hierarchy. Any additional effects (e.g. not allocating metadata) are up to the foo subsystem. This doesn't handle early_init subsystems (their "disabled" bit isn't set be, but it could easily be extended to do so if any of the early_init systems wanted it - I think it would just involve some nastier parameter processing since it would occur before the command-line argument parser had been run. Hugh said: Ballpark figures, I'm trying to get this question out rather than processing the exact numbers: CONFIG_CGROUP_MEM_RES_CTLR adds 15% overhead to the affected paths, booting with cgroup_disable=memory cuts that back to 1% overhead (due to slightly bigger struct page). I'm no expert on distros, they may have no interest whatever in CONFIG_CGROUP_MEM_RES_CTLR=y; and the rest of us can easily build with or without it, or apply the cgroup_disable=memory patches. Unix bench's execl test result on x86_64 was == just after boot without mounting any cgroup fs.== mem_cgorup=off : Execl Throughput 43.0 3150.1 732.6 mem_cgroup=on : Execl Throughput 43.0 2932.6 682.0 == [lizf@cn.fujitsu.com: fix boot option parsing] Signed-off-by: Balbir Singh Cc: Paul Menage Cc: Balbir Singh Cc: Pavel Emelyanov Cc: KAMEZAWA Hiroyuki Cc: Hugh Dickins Cc: Sudhir Kumar Cc: YAMAMOTO Takashi Cc: David Rientjes Signed-off-by: Li Zefan Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- Documentation/kernel-parameters.txt | 4 ++++ include/linux/cgroup.h | 1 + kernel/cgroup.c | 42 ++++++++++++++++++++++++++++++++----- 3 files changed, 42 insertions(+), 5 deletions(-) (limited to 'include/linux') diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 4cd1a5da80a4..32e9297ef747 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -375,6 +375,10 @@ and is between 256 and 4096 characters. It is defined in the file ccw_timeout_log [S390] See Documentation/s390/CommonIO for details. + cgroup_disable= [KNL] Disable a particular controller + Format: {name of the controller(s) to disable} + {Currently supported controllers - "memory"} + checkreqprot [SELINUX] Set initial checkreqprot flag value. Format: { "0" | "1" } See security/selinux/Kconfig help text. diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h index 028ba3b523b1..a6a6035a4e1e 100644 --- a/include/linux/cgroup.h +++ b/include/linux/cgroup.h @@ -256,6 +256,7 @@ struct cgroup_subsys { void (*bind)(struct cgroup_subsys *ss, struct cgroup *root); int subsys_id; int active; + int disabled; int early_init; #define MAX_CGROUP_TYPE_NAMELEN 32 const char *name; diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 53d86b4b0ce0..62f1a5231fe9 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -782,7 +782,14 @@ static int parse_cgroupfs_options(char *data, if (!*token) return -EINVAL; if (!strcmp(token, "all")) { - opts->subsys_bits = (1 << CGROUP_SUBSYS_COUNT) - 1; + /* Add all non-disabled subsystems */ + int i; + opts->subsys_bits = 0; + for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + struct cgroup_subsys *ss = subsys[i]; + if (!ss->disabled) + opts->subsys_bits |= 1ul << i; + } } else if (!strcmp(token, "noprefix")) { set_bit(ROOT_NOPREFIX, &opts->flags); } else if (!strncmp(token, "release_agent=", 14)) { @@ -800,7 +807,8 @@ static int parse_cgroupfs_options(char *data, for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { ss = subsys[i]; if (!strcmp(token, ss->name)) { - set_bit(i, &opts->subsys_bits); + if (!ss->disabled) + set_bit(i, &opts->subsys_bits); break; } } @@ -2600,13 +2608,13 @@ static int proc_cgroupstats_show(struct seq_file *m, void *v) { int i; - seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\n"); + seq_puts(m, "#subsys_name\thierarchy\tnum_cgroups\tenabled\n"); mutex_lock(&cgroup_mutex); for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { struct cgroup_subsys *ss = subsys[i]; - seq_printf(m, "%s\t%lu\t%d\n", + seq_printf(m, "%s\t%lu\t%d\t%d\n", ss->name, ss->root->subsys_bits, - ss->root->number_of_cgroups); + ss->root->number_of_cgroups, !ss->disabled); } mutex_unlock(&cgroup_mutex); return 0; @@ -3010,3 +3018,27 @@ static void cgroup_release_agent(struct work_struct *work) spin_unlock(&release_list_lock); mutex_unlock(&cgroup_mutex); } + +static int __init cgroup_disable(char *str) +{ + int i; + char *token; + + while ((token = strsep(&str, ",")) != NULL) { + if (!*token) + continue; + + for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) { + struct cgroup_subsys *ss = subsys[i]; + + if (!strcmp(token, ss->name)) { + ss->disabled = 1; + printk(KERN_INFO "Disabling %s control group" + " subsystem\n", ss->name); + break; + } + } + } + return 1; +} +__setup("cgroup_disable=", cgroup_disable); -- cgit v1.3 From 2557a933b795c1988c721ebb871cd735128bb9cb Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Mon, 7 Apr 2008 14:30:28 +1000 Subject: virtio: remove overzealous BUG_ON. The 'disable_cb' callback is designed as an optimization to tell the host we don't need callbacks now. As it is not reliable, the debug check is overzealous: it can happen on two CPUs at the same time. Document this. Even if it were reliable, the virtio_net driver doesn't disable callbacks on transmit so the START_USE/END_USE debugging reentrance protection can be easily tripped even on UP. Thanks to Balaji Rao for the bug report and testing. Signed-off-by: Rusty Russell CC: Balaji Rao Signed-off-by: Linus Torvalds --- drivers/virtio/virtio_ring.c | 3 --- include/linux/virtio.h | 5 ++++- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'include/linux') diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index aa714028641e..c2fa5c630813 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -214,10 +214,7 @@ static void vring_disable_cb(struct virtqueue *_vq) { struct vring_virtqueue *vq = to_vvq(_vq); - START_USE(vq); - BUG_ON(vq->vring.avail->flags & VRING_AVAIL_F_NO_INTERRUPT); vq->vring.avail->flags |= VRING_AVAIL_F_NO_INTERRUPT; - END_USE(vq); } static bool vring_enable_cb(struct virtqueue *_vq) diff --git a/include/linux/virtio.h b/include/linux/virtio.h index 12c18ac1b973..e7d10845b3c1 100644 --- a/include/linux/virtio.h +++ b/include/linux/virtio.h @@ -41,6 +41,8 @@ struct virtqueue * Returns NULL or the "data" token handed to add_buf. * @disable_cb: disable callbacks * vq: the struct virtqueue we're talking about. + * Note that this is not necessarily synchronous, hence unreliable and only + * useful as an optimization. * @enable_cb: restart callbacks after disable_cb. * vq: the struct virtqueue we're talking about. * This re-enables callbacks; it returns "false" if there are pending @@ -48,7 +50,8 @@ struct virtqueue * checking for more work, and enabling callbacks. * * Locking rules are straightforward: the driver is responsible for - * locking. No two operations may be invoked simultaneously. + * locking. No two operations may be invoked simultaneously, with the exception + * of @disable_cb. * * All operations can be called in any context. */ -- cgit v1.3 From 54a015104136974262afa4b8ddd943ea70dec8a2 Mon Sep 17 00:00:00 2001 From: Roland McGrath Date: Thu, 10 Apr 2008 15:37:38 -0700 Subject: asmlinkage_protect replaces prevent_tail_call The prevent_tail_call() macro works around the problem of the compiler clobbering argument words on the stack, which for asmlinkage functions is the caller's (user's) struct pt_regs. The tail/sibling-call optimization is not the only way that the compiler can decide to use stack argument words as scratch space, which we have to prevent. Other optimizations can do it too. Until we have new compiler support to make "asmlinkage" binding on the compiler's own use of the stack argument frame, we have work around all the manifestations of this issue that crop up. More cases seem to be prevented by also keeping the incoming argument variables live at the end of the function. This makes their original stack slots attractive places to leave those variables, so the compiler tends not clobber them for something else. It's still no guarantee, but it handles some observed cases that prevent_tail_call() did not. Signed-off-by: Roland McGrath Signed-off-by: Linus Torvalds --- arch/x86/kernel/tls.c | 4 ++-- fs/open.c | 8 ++++---- include/asm-x86/linkage.h | 24 +++++++++++++++++++++++- include/linux/linkage.h | 4 ++-- kernel/exit.c | 4 ++-- kernel/uid16.c | 22 +++++++++++----------- 6 files changed, 44 insertions(+), 22 deletions(-) (limited to 'include/linux') diff --git a/arch/x86/kernel/tls.c b/arch/x86/kernel/tls.c index 022bcaa3b42e..ab6bf375a307 100644 --- a/arch/x86/kernel/tls.c +++ b/arch/x86/kernel/tls.c @@ -92,7 +92,7 @@ int do_set_thread_area(struct task_struct *p, int idx, asmlinkage int sys_set_thread_area(struct user_desc __user *u_info) { int ret = do_set_thread_area(current, -1, u_info, 1); - prevent_tail_call(ret); + asmlinkage_protect(1, ret, u_info); return ret; } @@ -142,7 +142,7 @@ int do_get_thread_area(struct task_struct *p, int idx, asmlinkage int sys_get_thread_area(struct user_desc __user *u_info) { int ret = do_get_thread_area(current, -1, u_info); - prevent_tail_call(ret); + asmlinkage_protect(1, ret, u_info); return ret; } diff --git a/fs/open.c b/fs/open.c index a4b12022edaa..3fa4e4ffce4c 100644 --- a/fs/open.c +++ b/fs/open.c @@ -335,7 +335,7 @@ asmlinkage long sys_ftruncate(unsigned int fd, unsigned long length) { long ret = do_sys_ftruncate(fd, length, 1); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(2, ret, fd, length); return ret; } @@ -350,7 +350,7 @@ asmlinkage long sys_ftruncate64(unsigned int fd, loff_t length) { long ret = do_sys_ftruncate(fd, length, 0); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(2, ret, fd, length); return ret; } #endif @@ -1067,7 +1067,7 @@ asmlinkage long sys_open(const char __user *filename, int flags, int mode) ret = do_sys_open(AT_FDCWD, filename, flags, mode); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(3, ret, filename, flags, mode); return ret; } @@ -1081,7 +1081,7 @@ asmlinkage long sys_openat(int dfd, const char __user *filename, int flags, ret = do_sys_open(dfd, filename, flags, mode); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(4, ret, dfd, filename, flags, mode); return ret; } diff --git a/include/asm-x86/linkage.h b/include/asm-x86/linkage.h index 31739c7d66a9..d605eeba0f70 100644 --- a/include/asm-x86/linkage.h +++ b/include/asm-x86/linkage.h @@ -8,12 +8,34 @@ #ifdef CONFIG_X86_32 #define asmlinkage CPP_ASMLINKAGE __attribute__((regparm(0))) -#define prevent_tail_call(ret) __asm__ ("" : "=r" (ret) : "0" (ret)) /* * For 32-bit UML - mark functions implemented in assembly that use * regparm input parameters: */ #define asmregparm __attribute__((regparm(3))) + +#define asmlinkage_protect(n, ret, args...) \ + __asmlinkage_protect##n(ret, ##args) +#define __asmlinkage_protect_n(ret, args...) \ + __asm__ __volatile__ ("" : "=r" (ret) : "0" (ret), ##args) +#define __asmlinkage_protect0(ret) \ + __asmlinkage_protect_n(ret) +#define __asmlinkage_protect1(ret, arg1) \ + __asmlinkage_protect_n(ret, "g" (arg1)) +#define __asmlinkage_protect2(ret, arg1, arg2) \ + __asmlinkage_protect_n(ret, "g" (arg1), "g" (arg2)) +#define __asmlinkage_protect3(ret, arg1, arg2, arg3) \ + __asmlinkage_protect_n(ret, "g" (arg1), "g" (arg2), "g" (arg3)) +#define __asmlinkage_protect4(ret, arg1, arg2, arg3, arg4) \ + __asmlinkage_protect_n(ret, "g" (arg1), "g" (arg2), "g" (arg3), \ + "g" (arg4)) +#define __asmlinkage_protect5(ret, arg1, arg2, arg3, arg4, arg5) \ + __asmlinkage_protect_n(ret, "g" (arg1), "g" (arg2), "g" (arg3), \ + "g" (arg4), "g" (arg5)) +#define __asmlinkage_protect6(ret, arg1, arg2, arg3, arg4, arg5, arg6) \ + __asmlinkage_protect_n(ret, "g" (arg1), "g" (arg2), "g" (arg3), \ + "g" (arg4), "g" (arg5), "g" (arg6)) + #endif #ifdef CONFIG_X86_ALIGNMENT_16 diff --git a/include/linux/linkage.h b/include/linux/linkage.h index 0592936344c4..fe2a39c489b6 100644 --- a/include/linux/linkage.h +++ b/include/linux/linkage.h @@ -17,8 +17,8 @@ # define asmregparm #endif -#ifndef prevent_tail_call -# define prevent_tail_call(ret) do { } while (0) +#ifndef asmlinkage_protect +# define asmlinkage_protect(n, ret, args...) do { } while (0) #endif #ifndef __ALIGN diff --git a/kernel/exit.c b/kernel/exit.c index 53872bf993fa..073005b1cfb2 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -1608,7 +1608,7 @@ asmlinkage long sys_waitid(int which, pid_t upid, put_pid(pid); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(5, ret, which, upid, infop, options, ru); return ret; } @@ -1640,7 +1640,7 @@ asmlinkage long sys_wait4(pid_t upid, int __user *stat_addr, put_pid(pid); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(4, ret, upid, stat_addr, options, ru); return ret; } diff --git a/kernel/uid16.c b/kernel/uid16.c index dd308ba4e03b..3e41c1673e2f 100644 --- a/kernel/uid16.c +++ b/kernel/uid16.c @@ -21,7 +21,7 @@ asmlinkage long sys_chown16(const char __user * filename, old_uid_t user, old_gi { long ret = sys_chown(filename, low2highuid(user), low2highgid(group)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(3, ret, filename, user, group); return ret; } @@ -29,7 +29,7 @@ asmlinkage long sys_lchown16(const char __user * filename, old_uid_t user, old_g { long ret = sys_lchown(filename, low2highuid(user), low2highgid(group)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(3, ret, filename, user, group); return ret; } @@ -37,7 +37,7 @@ asmlinkage long sys_fchown16(unsigned int fd, old_uid_t user, old_gid_t group) { long ret = sys_fchown(fd, low2highuid(user), low2highgid(group)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(3, ret, fd, user, group); return ret; } @@ -45,7 +45,7 @@ asmlinkage long sys_setregid16(old_gid_t rgid, old_gid_t egid) { long ret = sys_setregid(low2highgid(rgid), low2highgid(egid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(2, ret, rgid, egid); return ret; } @@ -53,7 +53,7 @@ asmlinkage long sys_setgid16(old_gid_t gid) { long ret = sys_setgid(low2highgid(gid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(1, ret, gid); return ret; } @@ -61,7 +61,7 @@ asmlinkage long sys_setreuid16(old_uid_t ruid, old_uid_t euid) { long ret = sys_setreuid(low2highuid(ruid), low2highuid(euid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(2, ret, ruid, euid); return ret; } @@ -69,7 +69,7 @@ asmlinkage long sys_setuid16(old_uid_t uid) { long ret = sys_setuid(low2highuid(uid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(1, ret, uid); return ret; } @@ -78,7 +78,7 @@ asmlinkage long sys_setresuid16(old_uid_t ruid, old_uid_t euid, old_uid_t suid) long ret = sys_setresuid(low2highuid(ruid), low2highuid(euid), low2highuid(suid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(3, ret, ruid, euid, suid); return ret; } @@ -98,7 +98,7 @@ asmlinkage long sys_setresgid16(old_gid_t rgid, old_gid_t egid, old_gid_t sgid) long ret = sys_setresgid(low2highgid(rgid), low2highgid(egid), low2highgid(sgid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(3, ret, rgid, egid, sgid); return ret; } @@ -117,7 +117,7 @@ asmlinkage long sys_setfsuid16(old_uid_t uid) { long ret = sys_setfsuid(low2highuid(uid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(1, ret, uid); return ret; } @@ -125,7 +125,7 @@ asmlinkage long sys_setfsgid16(old_gid_t gid) { long ret = sys_setfsgid(low2highgid(gid)); /* avoid REGPARM breakage on x86: */ - prevent_tail_call(ret); + asmlinkage_protect(1, ret, gid); return ret; } -- cgit v1.3 From d10d89ec78114f925f63c5126a2b2490f501a462 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Thu, 10 Apr 2008 17:35:23 -0700 Subject: Add commentary about the new "asmlinkage_protect()" macro It's really a pretty ugly thing to need, and some day it will hopefully be obviated by teaching gcc about the magic calling conventions for the low-level system call code, but in the meantime we can at least add big honking comments about why we need these insane and strange macros. I took my comments from my version of the macro, but I ended up deciding to just pick Roland's version of the actual code instead (with his prettier syntax that uses vararg macros). Thus the previous two commits that actually implement it. Signed-off-by: Linus Torvalds --- include/asm-x86/linkage.h | 11 +++++++++++ include/linux/linkage.h | 13 +++++++++++++ 2 files changed, 24 insertions(+) (limited to 'include/linux') diff --git a/include/asm-x86/linkage.h b/include/asm-x86/linkage.h index d605eeba0f70..c048353f4b85 100644 --- a/include/asm-x86/linkage.h +++ b/include/asm-x86/linkage.h @@ -14,6 +14,17 @@ */ #define asmregparm __attribute__((regparm(3))) +/* + * Make sure the compiler doesn't do anything stupid with the + * arguments on the stack - they are owned by the *caller*, not + * the callee. This just fools gcc into not spilling into them, + * and keeps it from doing tailcall recursion and/or using the + * stack slots for temporaries, since they are live and "used" + * all the way to the end of the function. + * + * NOTE! On x86-64, all the arguments are in registers, so this + * only matters on a 32-bit kernel. + */ #define asmlinkage_protect(n, ret, args...) \ __asmlinkage_protect##n(ret, ##args) #define __asmlinkage_protect_n(ret, args...) \ diff --git a/include/linux/linkage.h b/include/linux/linkage.h index fe2a39c489b6..b163c5c40dbc 100644 --- a/include/linux/linkage.h +++ b/include/linux/linkage.h @@ -17,6 +17,19 @@ # define asmregparm #endif +/* + * This is used by architectures to keep arguments on the stack + * untouched by the compiler by keeping them live until the end. + * The argument stack may be owned by the assembly-language + * caller, not the callee, and gcc doesn't always understand + * that. + * + * We have the return value, and a maximum of six arguments. + * + * This should always be followed by a "return ret" for the + * protection to work (ie no more work that the compiler might + * end up needing stack temporaries for). + */ #ifndef asmlinkage_protect # define asmlinkage_protect(n, ret, args...) do { } while (0) #endif -- cgit v1.3 From 544451a1a36b06f43fc67112ede3b92e6203eb9d Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 10 Apr 2008 21:29:28 -0700 Subject: pnp: increase number of devices supported per protocol Increase the PNP "number of devices" limit. We currently use an unsigned char, which limits us to 256 devices per protocol. This patch changes that to an unsigned int. Not all backends can take advantage of this: we limit ISAPNP to 10 devices in isapnp_cfg_begin(), and PNPBIOS is limited to 256 devices because the BIOS interfaces use a one-byte device node number. But there is no limit on the number of PNPACPI devices we may have. Large HP Integrity machines have more than 256, which causes the current "unsigned char number" to wrap around. This causes errors like this: pnp: PnP ACPI init kobject_add failed for 00:00 with -EEXIST, don't try to register things with the same name in the same directory. Call Trace: [] show_stack+0x40/0xa0 [] dump_stack+0x30/0x60 [] kobject_add+0x290/0x2c0 [] device_add+0x160/0x860 [] device_register+0x30/0x60 [] __pnp_add_device+0x130/0x180 [] pnp_add_device+0xb0/0xe0 [] pnpacpi_add_device+0x510/0x5a0 [] pnpacpi_add_device_handler+0x50/0x80 This patch increases the limit to fix this PNPACPI problem. It should not have any adverse effect on ISAPNP or PNPBIOS because their limits are still enforced in the backends. Signed-off-by: Bjorn Helgaas Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/pnp.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/pnp.h b/include/linux/pnp.h index 29dd55838e84..b2f05c230f4b 100644 --- a/include/linux/pnp.h +++ b/include/linux/pnp.h @@ -175,7 +175,7 @@ static inline void pnp_set_card_drvdata(struct pnp_card_link *pcard, void *data) struct pnp_dev { struct device dev; /* Driver Model device interface */ u64 dma_mask; - unsigned char number; /* used as an index, must be unique */ + unsigned int number; /* used as an index, must be unique */ int status; struct list_head global_list; /* node in global list of devices */ -- cgit v1.3 From b0fac02370cffad956ff3de5e8ed4df7e7b875d7 Mon Sep 17 00:00:00 2001 From: Heiko Carstens Date: Fri, 11 Apr 2008 13:46:54 +0200 Subject: Fix "$(AS) -traditional" compile breakage caused by asmlinkage_protect git commit 54a015104136974262afa4b8ddd943ea70dec8a2 ("asmlinkage_protect replaces prevent_tail_call") causes this build failure on s390: AS arch/s390/kernel/entry64.o In file included from arch/s390/kernel/entry64.S:14: include/linux/linkage.h:34: error: syntax error in macro parameter list make[1]: *** [arch/s390/kernel/entry64.o] Error 1 make: *** [arch/s390/kernel] Error 2 and some other architectures. The reason is that some architectures add the "-traditional" flag to the invocation of $(AS), which disables variadic macro argument support. So just surround the new define with an #ifndef __ASSEMBLY__ to prevent any side effects on asm code. Cc: Roland McGrath Cc: Martin Schwidefsky Signed-off-by: Heiko Carstens Signed-off-by: Linus Torvalds --- include/linux/linkage.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/linkage.h b/include/linux/linkage.h index b163c5c40dbc..2119610b24f8 100644 --- a/include/linux/linkage.h +++ b/include/linux/linkage.h @@ -30,9 +30,12 @@ * protection to work (ie no more work that the compiler might * end up needing stack temporaries for). */ +/* Assembly files may be compiled with -traditional .. */ +#ifndef __ASSEMBLY__ #ifndef asmlinkage_protect # define asmlinkage_protect(n, ret, args...) do { } while (0) #endif +#endif #ifndef __ALIGN #define __ALIGN .align 4,0x90 -- cgit v1.3 From dc07e721a26ec7e0adb66340f1763d220cfbbd0c Mon Sep 17 00:00:00 2001 From: "J. Bruce Fields" Date: Mon, 7 Apr 2008 15:59:05 -0400 Subject: Spell out behavior of atomic_dec_and_lock() in kerneldoc A little more detail here wouldn't hurt. Signed-off-by: J. Bruce Fields Signed-off-by: Jonathan Corbet --- include/linux/spinlock.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/spinlock.h b/include/linux/spinlock.h index 576a5f77d3bd..1129ee0a7180 100644 --- a/include/linux/spinlock.h +++ b/include/linux/spinlock.h @@ -341,6 +341,9 @@ static inline void double_spin_unlock(spinlock_t *l1, spinlock_t *l2, * atomic_dec_and_lock - lock on reaching reference count zero * @atomic: the atomic counter * @lock: the spinlock in question + * + * Decrements @atomic by 1. If the result is 0, returns true and locks + * @lock. Returns false for all other cases. */ extern int _atomic_dec_and_lock(atomic_t *atomic, spinlock_t *lock); #define atomic_dec_and_lock(atomic, lock) \ -- cgit v1.3