From 979f693def9084a452846365dfde5dcb28366333 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 22 Sep 2009 14:44:11 +0200 Subject: ratelimit: Use per ratelimit context locking I'd like to use printk_ratelimit() in atomic context, but that's not possible right now due to the spinlock usage this commit introduced more than a year ago: 717115e: printk ratelimiting rewrite As a first step push the lock into the ratelimit state structure. This allows us to deal with locking failures to be considered as an event related to that state being too busy. Also clean up the code a bit (without changing functionality): - tidy up the definitions - clean up the code flow This also shrinks the code a tiny bit: text data bss dec hex filename 264 0 4 268 10c ratelimit.o.before 255 0 0 255 ff ratelimit.o.after ( Whole-kernel data size got a bit larger, because we have two ratelimit-state data structures right now. ) Cc: Peter Zijlstra Cc: Andrew Morton Cc: Linus Torvalds Cc: David S. Miller LKML-Reference: Signed-off-by: Ingo Molnar --- lib/ratelimit.c | 29 +++++++++++++---------------- 1 file changed, 13 insertions(+), 16 deletions(-) (limited to 'lib') diff --git a/lib/ratelimit.c b/lib/ratelimit.c index 26187edcc7ea..0e2c28e8a0ca 100644 --- a/lib/ratelimit.c +++ b/lib/ratelimit.c @@ -7,15 +7,12 @@ * parameter. Now every user can use their own standalone ratelimit_state. * * This file is released under the GPLv2. - * */ #include #include #include -static DEFINE_SPINLOCK(ratelimit_lock); - /* * __ratelimit - rate limiting * @rs: ratelimit_state data @@ -26,11 +23,12 @@ static DEFINE_SPINLOCK(ratelimit_lock); int __ratelimit(struct ratelimit_state *rs) { unsigned long flags; + int ret; if (!rs->interval) return 1; - spin_lock_irqsave(&ratelimit_lock, flags); + spin_lock_irqsave(&rs->lock, flags); if (!rs->begin) rs->begin = jiffies; @@ -38,20 +36,19 @@ int __ratelimit(struct ratelimit_state *rs) if (rs->missed) printk(KERN_WARNING "%s: %d callbacks suppressed\n", __func__, rs->missed); - rs->begin = 0; + rs->begin = 0; rs->printed = 0; - rs->missed = 0; + rs->missed = 0; } - if (rs->burst && rs->burst > rs->printed) - goto print; - - rs->missed++; - spin_unlock_irqrestore(&ratelimit_lock, flags); - return 0; + if (rs->burst && rs->burst > rs->printed) { + rs->printed++; + ret = 1; + } else { + rs->missed++; + ret = 0; + } + spin_unlock_irqrestore(&rs->lock, flags); -print: - rs->printed++; - spin_unlock_irqrestore(&ratelimit_lock, flags); - return 1; + return ret; } EXPORT_SYMBOL(__ratelimit); -- cgit v1.2.3 From edaac8e3167501cda336231d00611bf59c164346 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 22 Sep 2009 14:44:11 +0200 Subject: ratelimit: Fix/allow use in atomic contexts I'd like to use printk_ratelimit() in NMI context, but it's not robust right now due to spinlock usage in lib/ratelimit.c. If an NMI is unlucky enough to hit just that spot we might lock up trying to take the spinlock again. Fix that by using a trylock variant. If we contend on that lock we can genuinely skip the message because the state is just being accessed by another CPU (or by this CPU). ( We could use atomics for the suppressed messages field, but i doubt it matters in practice and it makes the code heavier. ) Cc: Peter Zijlstra Cc: Andrew Morton Cc: Linus Torvalds Cc: David S. Miller LKML-Reference: Signed-off-by: Ingo Molnar --- lib/ratelimit.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) (limited to 'lib') diff --git a/lib/ratelimit.c b/lib/ratelimit.c index 0e2c28e8a0ca..69bfcacda16d 100644 --- a/lib/ratelimit.c +++ b/lib/ratelimit.c @@ -28,7 +28,15 @@ int __ratelimit(struct ratelimit_state *rs) if (!rs->interval) return 1; - spin_lock_irqsave(&rs->lock, flags); + /* + * If we contend on this state's lock then almost + * by definition we are too busy to print a message, + * in addition to the one that will be printed by + * the entity that is holding the lock already: + */ + if (!spin_trylock_irqsave(&rs->lock, flags)) + return 1; + if (!rs->begin) rs->begin = jiffies; -- cgit v1.2.3 From 3fff4c42bd0a89869a0eb1e7874cc06ffa4aa0f5 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Tue, 22 Sep 2009 16:18:09 +0200 Subject: printk: Remove ratelimit.h from kernel.h Decouple kernel.h from ratelimit.h: the global declaration of printk's ratelimit_state is not needed, and it leads to messy circular dependencies due to ratelimit.h's (new) adding of a spinlock_types.h include. Cc: Peter Zijlstra Cc: Andrew Morton Cc: Linus Torvalds Cc: David S. Miller LKML-Reference: Signed-off-by: Ingo Molnar --- include/linux/kernel.h | 2 -- include/linux/net.h | 1 + kernel/printk.c | 1 + kernel/sysctl.c | 3 +++ lib/ratelimit.c | 2 +- net/core/sysctl_net_core.c | 2 ++ net/core/utils.c | 2 ++ 7 files changed, 10 insertions(+), 3 deletions(-) (limited to 'lib') diff --git a/include/linux/kernel.h b/include/linux/kernel.h index 2b5b1e0899a8..3305f33201be 100644 --- a/include/linux/kernel.h +++ b/include/linux/kernel.h @@ -15,7 +15,6 @@ #include #include #include -#include #include #include #include @@ -241,7 +240,6 @@ asmlinkage int vprintk(const char *fmt, va_list args) asmlinkage int printk(const char * fmt, ...) __attribute__ ((format (printf, 1, 2))) __cold; -extern struct ratelimit_state printk_ratelimit_state; extern int printk_ratelimit(void); extern bool printk_timed_ratelimit(unsigned long *caller_jiffies, unsigned int interval_msec); diff --git a/include/linux/net.h b/include/linux/net.h index 9040a10584f7..df20f680f455 100644 --- a/include/linux/net.h +++ b/include/linux/net.h @@ -358,6 +358,7 @@ static const struct proto_ops name##_ops = { \ #ifdef CONFIG_SYSCTL #include +#include extern struct ratelimit_state net_ratelimit_state; #endif diff --git a/kernel/printk.c b/kernel/printk.c index 602033acd6c7..b997c893cdcf 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -33,6 +33,7 @@ #include #include #include +#include #include diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 1a631ba684a4..6c37048b9db9 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include #include @@ -155,6 +156,8 @@ extern int no_unaligned_warning; extern int unaligned_dump_stack; #endif +extern struct ratelimit_state printk_ratelimit_state; + #ifdef CONFIG_RT_MUTEXES extern int max_lock_depth; #endif diff --git a/lib/ratelimit.c b/lib/ratelimit.c index 69bfcacda16d..5551731ae1d4 100644 --- a/lib/ratelimit.c +++ b/lib/ratelimit.c @@ -9,7 +9,7 @@ * This file is released under the GPLv2. */ -#include +#include #include #include diff --git a/net/core/sysctl_net_core.c b/net/core/sysctl_net_core.c index 7db1de0497c6..887c03c4e3c6 100644 --- a/net/core/sysctl_net_core.c +++ b/net/core/sysctl_net_core.c @@ -10,7 +10,9 @@ #include #include #include +#include #include + #include #include diff --git a/net/core/utils.c b/net/core/utils.c index 83221aee7084..838250241d26 100644 --- a/net/core/utils.c +++ b/net/core/utils.c @@ -24,6 +24,8 @@ #include #include #include +#include + #include #include -- cgit v1.2.3 From 96a2c464de07d7c72988db851c029b204fc59108 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Sat, 1 Aug 2009 01:34:24 +0200 Subject: tracing/bkl: Add bkl ftrace events Add two events lock_kernel and unlock_kernel() to trace the bkl uses. This opens the door for userspace tools to perform statistics about the callsites that use it, dependencies with other locks (by pairing the trace with lock events), use with recursivity and so on... The {__reacquire,release}_kernel_lock() events are not traced because these are called from schedule, thus the sched events are sufficient to trace them. Example of a trace: hald-addon-stor-4152 [000] 165.875501: unlock_kernel: depth: 0, fs/block_dev.c:1358 __blkdev_put() hald-addon-stor-4152 [000] 167.832974: lock_kernel: depth: 0, fs/block_dev.c:1167 __blkdev_get() How to get the callsites that acquire it recursively: cd /debug/tracing/events/bkl echo "lock_depth > 0" > filter firefox-4951 [001] 206.276967: unlock_kernel: depth: 1, fs/reiserfs/super.c:575 reiserfs_dirty_inode() You can also filter by file and/or line. v2: Use of FILTER_PTR_STRING attribute for files and lines fields to make them traceable. Signed-off-by: Frederic Weisbecker Cc: Steven Rostedt Cc: Li Zefan --- include/linux/smp_lock.h | 19 ++++++++++++--- include/trace/events/bkl.h | 61 ++++++++++++++++++++++++++++++++++++++++++++++ lib/kernel_lock.c | 11 +++++---- 3 files changed, 82 insertions(+), 9 deletions(-) create mode 100644 include/trace/events/bkl.h (limited to 'lib') diff --git a/include/linux/smp_lock.h b/include/linux/smp_lock.h index 813be59bf345..d48cc77ba70d 100644 --- a/include/linux/smp_lock.h +++ b/include/linux/smp_lock.h @@ -3,6 +3,7 @@ #ifdef CONFIG_LOCK_KERNEL #include +#include #define kernel_locked() (current->lock_depth >= 0) @@ -24,8 +25,18 @@ static inline int reacquire_kernel_lock(struct task_struct *task) return 0; } -extern void __lockfunc lock_kernel(void) __acquires(kernel_lock); -extern void __lockfunc unlock_kernel(void) __releases(kernel_lock); +extern void __lockfunc _lock_kernel(void) __acquires(kernel_lock); +extern void __lockfunc _unlock_kernel(void) __releases(kernel_lock); + +#define lock_kernel() { \ + trace_lock_kernel(__func__, __FILE__, __LINE__); \ + _lock_kernel(); \ +} + +#define unlock_kernel() { \ + trace_unlock_kernel(__func__, __FILE__, __LINE__); \ + _unlock_kernel(); \ +} /* * Various legacy drivers don't really need the BKL in a specific @@ -41,8 +52,8 @@ static inline void cycle_kernel_lock(void) #else -#define lock_kernel() do { } while(0) -#define unlock_kernel() do { } while(0) +#define lock_kernel() trace_lock_kernel(__func__, __FILE__, __LINE__); +#define unlock_kernel() trace_unlock_kernel(__func__, __FILE__, __LINE__); #define release_kernel_lock(task) do { } while(0) #define cycle_kernel_lock() do { } while(0) #define reacquire_kernel_lock(task) 0 diff --git a/include/trace/events/bkl.h b/include/trace/events/bkl.h new file mode 100644 index 000000000000..8abd620a490e --- /dev/null +++ b/include/trace/events/bkl.h @@ -0,0 +1,61 @@ +#undef TRACE_SYSTEM +#define TRACE_SYSTEM bkl + +#if !defined(_TRACE_BKL_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_BKL_H + +#include + +TRACE_EVENT(lock_kernel, + + TP_PROTO(const char *func, const char *file, int line), + + TP_ARGS(func, file, line), + + TP_STRUCT__entry( + __field( int, lock_depth ) + __field_ext( const char *, func, FILTER_PTR_STRING ) + __field_ext( const char *, file, FILTER_PTR_STRING ) + __field( int, line ) + ), + + TP_fast_assign( + /* We want to record the lock_depth after lock is acquired */ + __entry->lock_depth = current->lock_depth + 1; + __entry->func = func; + __entry->file = file; + __entry->line = line; + ), + + TP_printk("depth: %d, %s:%d %s()", __entry->lock_depth, + __entry->file, __entry->line, __entry->func) +); + +TRACE_EVENT(unlock_kernel, + + TP_PROTO(const char *func, const char *file, int line), + + TP_ARGS(func, file, line), + + TP_STRUCT__entry( + __field(int, lock_depth) + __field(const char *, func) + __field(const char *, file) + __field(int, line) + ), + + TP_fast_assign( + __entry->lock_depth = current->lock_depth; + __entry->func = func; + __entry->file = file; + __entry->line = line; + ), + + TP_printk("depth: %d, %s:%d %s()", __entry->lock_depth, + __entry->file, __entry->line, __entry->func) +); + +#endif /* _TRACE_BKL_H */ + +/* This part must be outside protection */ +#include diff --git a/lib/kernel_lock.c b/lib/kernel_lock.c index 39f1029e3525..5c10b2e1fd08 100644 --- a/lib/kernel_lock.c +++ b/lib/kernel_lock.c @@ -5,10 +5,11 @@ * relegated to obsolescence, but used by various less * important (or lazy) subsystems. */ -#include #include #include #include +#define CREATE_TRACE_POINTS +#include /* * The 'big kernel lock' @@ -113,7 +114,7 @@ static inline void __unlock_kernel(void) * This cannot happen asynchronously, so we only need to * worry about other CPU's. */ -void __lockfunc lock_kernel(void) +void __lockfunc _lock_kernel(void) { int depth = current->lock_depth+1; if (likely(!depth)) @@ -121,13 +122,13 @@ void __lockfunc lock_kernel(void) current->lock_depth = depth; } -void __lockfunc unlock_kernel(void) +void __lockfunc _unlock_kernel(void) { BUG_ON(current->lock_depth < 0); if (likely(--current->lock_depth < 0)) __unlock_kernel(); } -EXPORT_SYMBOL(lock_kernel); -EXPORT_SYMBOL(unlock_kernel); +EXPORT_SYMBOL(_lock_kernel); +EXPORT_SYMBOL(_unlock_kernel); -- cgit v1.2.3 From 925936ebf35a95c290e010b784c962164e6728f3 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Mon, 28 Sep 2009 17:12:49 +0200 Subject: tracing: Pushdown the bkl tracepoints calls Currently we are calling the bkl tracepoint callbacks just before the bkl lock/unlock operations, ie the tracepoint call is not inside a lock_kernel() function but inside a lock_kernel() macro. Hence the bkl trace event header must be included from smp_lock.h. This raises some nasty circular header dependencies: linux/smp_lock.h -> trace/events/bkl.h -> trace/define_trace.h -> trace/ftrace.h -> linux/ftrace_event.h -> linux/hardirq.h -> linux/smp_lock.h This results in incomplete event declarations, spurious event definitions and other kind of funny behaviours. This is hardly fixable without ugly workarounds. So instead, we push the file name, line number and function name as lock_kernel() parameters, so that we only deal with the trace event header from lib/kernel_lock.c This adds two parameters to lock_kernel() and unlock_kernel() but it should be fine wrt to performances because this pair dos not seem to be called in fast paths. Signed-off-by: Frederic Weisbecker Cc: Steven Rostedt Cc: Ingo Molnar Cc: Li Zefan --- include/linux/smp_lock.h | 28 +++++++++++++++------------- lib/kernel_lock.c | 15 +++++++++++---- 2 files changed, 26 insertions(+), 17 deletions(-) (limited to 'lib') diff --git a/include/linux/smp_lock.h b/include/linux/smp_lock.h index d48cc77ba70d..2ea1dd1ba21c 100644 --- a/include/linux/smp_lock.h +++ b/include/linux/smp_lock.h @@ -3,7 +3,6 @@ #ifdef CONFIG_LOCK_KERNEL #include -#include #define kernel_locked() (current->lock_depth >= 0) @@ -25,18 +24,21 @@ static inline int reacquire_kernel_lock(struct task_struct *task) return 0; } -extern void __lockfunc _lock_kernel(void) __acquires(kernel_lock); -extern void __lockfunc _unlock_kernel(void) __releases(kernel_lock); +extern void __lockfunc +_lock_kernel(const char *func, const char *file, int line) +__acquires(kernel_lock); -#define lock_kernel() { \ - trace_lock_kernel(__func__, __FILE__, __LINE__); \ - _lock_kernel(); \ -} +extern void __lockfunc +_unlock_kernel(const char *func, const char *file, int line) +__releases(kernel_lock); -#define unlock_kernel() { \ - trace_unlock_kernel(__func__, __FILE__, __LINE__); \ - _unlock_kernel(); \ -} +#define lock_kernel() do { \ + _lock_kernel(__func__, __FILE__, __LINE__); \ +} while (0) + +#define unlock_kernel() do { \ + _unlock_kernel(__func__, __FILE__, __LINE__); \ +} while (0) /* * Various legacy drivers don't really need the BKL in a specific @@ -52,8 +54,8 @@ static inline void cycle_kernel_lock(void) #else -#define lock_kernel() trace_lock_kernel(__func__, __FILE__, __LINE__); -#define unlock_kernel() trace_unlock_kernel(__func__, __FILE__, __LINE__); +#define lock_kernel() +#define unlock_kernel() #define release_kernel_lock(task) do { } while(0) #define cycle_kernel_lock() do { } while(0) #define reacquire_kernel_lock(task) 0 diff --git a/lib/kernel_lock.c b/lib/kernel_lock.c index 5c10b2e1fd08..4ebfa5a164d7 100644 --- a/lib/kernel_lock.c +++ b/lib/kernel_lock.c @@ -8,9 +8,11 @@ #include #include #include -#define CREATE_TRACE_POINTS #include +#define CREATE_TRACE_POINTS +#include + /* * The 'big kernel lock' * @@ -114,19 +116,24 @@ static inline void __unlock_kernel(void) * This cannot happen asynchronously, so we only need to * worry about other CPU's. */ -void __lockfunc _lock_kernel(void) +void __lockfunc _lock_kernel(const char *func, const char *file, int line) { - int depth = current->lock_depth+1; + int depth = current->lock_depth + 1; + + trace_lock_kernel(func, file, line); + if (likely(!depth)) __lock_kernel(); current->lock_depth = depth; } -void __lockfunc _unlock_kernel(void) +void __lockfunc _unlock_kernel(const char *func, const char *file, int line) { BUG_ON(current->lock_depth < 0); if (likely(--current->lock_depth < 0)) __unlock_kernel(); + + trace_unlock_kernel(func, file, line); } EXPORT_SYMBOL(_lock_kernel); -- cgit v1.2.3 From 5c828713358cb9df8aa174371edcbbb62203a490 Mon Sep 17 00:00:00 2001 From: Christian Borntraeger Date: Fri, 23 Oct 2009 14:58:11 +0200 Subject: ratelimit: Make suppressed output messages more useful Today I got: [39648.224782] Registered led device: iwl-phy0::TX [40676.545099] __ratelimit: 246 callbacks suppressed [40676.545103] abcdef[23675]: segfault at 0 ... as you can see the ratelimit message contains a function prefix. Since this is always __ratelimit, this wont help much. This patch changes __ratelimit and printk_ratelimit to print the function name that calls ratelimit. This will pinpoint the responsible function, as long as not several different places call ratelimit with the same ratelimit state at the same time. In that case we catch only one random function that calls ratelimit after the wait period. Signed-off-by: Christian Borntraeger Cc: Dave Young Cc: Linus Torvalds CC: Andrew Morton LKML-Reference: <200910231458.11832.borntraeger@de.ibm.com> Signed-off-by: Ingo Molnar --- include/linux/kernel.h | 3 ++- include/linux/ratelimit.h | 3 ++- kernel/printk.c | 6 +++--- lib/ratelimit.c | 6 +++--- 4 files changed, 10 insertions(+), 8 deletions(-) (limited to 'lib') diff --git a/include/linux/kernel.h b/include/linux/kernel.h index 3305f33201be..21d0d822c716 100644 --- a/include/linux/kernel.h +++ b/include/linux/kernel.h @@ -240,7 +240,8 @@ asmlinkage int vprintk(const char *fmt, va_list args) asmlinkage int printk(const char * fmt, ...) __attribute__ ((format (printf, 1, 2))) __cold; -extern int printk_ratelimit(void); +extern int __printk_ratelimit(const char *func); +#define printk_ratelimit() __printk_ratelimit(__func__) extern bool printk_timed_ratelimit(unsigned long *caller_jiffies, unsigned int interval_msec); diff --git a/include/linux/ratelimit.h b/include/linux/ratelimit.h index 187bc16c1f15..668cf1bef030 100644 --- a/include/linux/ratelimit.h +++ b/include/linux/ratelimit.h @@ -25,6 +25,7 @@ struct ratelimit_state { .burst = burst_init, \ } -extern int __ratelimit(struct ratelimit_state *rs); +extern int ___ratelimit(struct ratelimit_state *rs, const char *func); +#define __ratelimit(state) ___ratelimit(state, __func__) #endif /* _LINUX_RATELIMIT_H */ diff --git a/kernel/printk.c b/kernel/printk.c index b997c893cdcf..8283dbe15af4 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1364,11 +1364,11 @@ late_initcall(disable_boot_consoles); */ DEFINE_RATELIMIT_STATE(printk_ratelimit_state, 5 * HZ, 10); -int printk_ratelimit(void) +int __printk_ratelimit(const char *func) { - return __ratelimit(&printk_ratelimit_state); + return ___ratelimit(&printk_ratelimit_state, func); } -EXPORT_SYMBOL(printk_ratelimit); +EXPORT_SYMBOL(__printk_ratelimit); /** * printk_timed_ratelimit - caller-controlled printk ratelimiting diff --git a/lib/ratelimit.c b/lib/ratelimit.c index 5551731ae1d4..09f5ce1810dc 100644 --- a/lib/ratelimit.c +++ b/lib/ratelimit.c @@ -20,7 +20,7 @@ * This enforces a rate limit: not more than @rs->ratelimit_burst callbacks * in every @rs->ratelimit_jiffies */ -int __ratelimit(struct ratelimit_state *rs) +int ___ratelimit(struct ratelimit_state *rs, const char *func) { unsigned long flags; int ret; @@ -43,7 +43,7 @@ int __ratelimit(struct ratelimit_state *rs) if (time_is_before_jiffies(rs->begin + rs->interval)) { if (rs->missed) printk(KERN_WARNING "%s: %d callbacks suppressed\n", - __func__, rs->missed); + func, rs->missed); rs->begin = 0; rs->printed = 0; rs->missed = 0; @@ -59,4 +59,4 @@ int __ratelimit(struct ratelimit_state *rs) return ret; } -EXPORT_SYMBOL(__ratelimit); +EXPORT_SYMBOL(___ratelimit); -- cgit v1.2.3 From 5740afdb68abadc473fd5392df733558a58c1254 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Tue, 10 Nov 2009 19:46:18 +0900 Subject: swiotlb: Add swiotlb_free() function swiotlb_free() function frees all allocated memory for swiotlb. We need to initialize swiotlb before IOMMU initialization (x86 and powerpc needs to allocate memory from bootmem allocator). If IOMMU initialization is successful, we need to free swiotlb resource (don't want to waste 64MB). Signed-off-by: FUJITA Tomonori Cc: chrisw@sous-sol.org Cc: dwmw2@infradead.org Cc: joerg.roedel@amd.com Cc: muli@il.ibm.com LKML-Reference: <1257849980-22640-8-git-send-email-fujita.tomonori@lab.ntt.co.jp> [ -v2: build fix for the !CONFIG_SWIOTLB case ] Signed-off-by: Ingo Molnar --- include/linux/swiotlb.h | 6 ++++++ lib/swiotlb.c | 30 ++++++++++++++++++++++++++++++ 2 files changed, 36 insertions(+) (limited to 'lib') diff --git a/include/linux/swiotlb.h b/include/linux/swiotlb.h index 73b1f1cec423..59bafa690290 100644 --- a/include/linux/swiotlb.h +++ b/include/linux/swiotlb.h @@ -88,4 +88,10 @@ swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr); extern int swiotlb_dma_supported(struct device *hwdev, u64 mask); +#ifdef CONFIG_SWIOTLB +extern void __init swiotlb_free(void); +#else +static inline void swiotlb_free(void) { } +#endif + #endif /* __LINUX_SWIOTLB_H */ diff --git a/lib/swiotlb.c b/lib/swiotlb.c index ac25cd28e807..eee512b63f17 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -97,6 +97,8 @@ static phys_addr_t *io_tlb_orig_addr; */ static DEFINE_SPINLOCK(io_tlb_lock); +static int late_alloc; + static int __init setup_io_tlb_npages(char *str) { @@ -262,6 +264,8 @@ swiotlb_late_init_with_default_size(size_t default_size) swiotlb_print_info(bytes); + late_alloc = 1; + return 0; cleanup4: @@ -281,6 +285,32 @@ cleanup1: return -ENOMEM; } +void __init swiotlb_free(void) +{ + if (!io_tlb_overflow_buffer) + return; + + if (late_alloc) { + free_pages((unsigned long)io_tlb_overflow_buffer, + get_order(io_tlb_overflow)); + free_pages((unsigned long)io_tlb_orig_addr, + get_order(io_tlb_nslabs * sizeof(phys_addr_t))); + free_pages((unsigned long)io_tlb_list, get_order(io_tlb_nslabs * + sizeof(int))); + free_pages((unsigned long)io_tlb_start, + get_order(io_tlb_nslabs << IO_TLB_SHIFT)); + } else { + free_bootmem_late(__pa(io_tlb_overflow_buffer), + io_tlb_overflow); + free_bootmem_late(__pa(io_tlb_orig_addr), + io_tlb_nslabs * sizeof(phys_addr_t)); + free_bootmem_late(__pa(io_tlb_list), + io_tlb_nslabs * sizeof(int)); + free_bootmem_late(__pa(io_tlb_start), + io_tlb_nslabs << IO_TLB_SHIFT); + } +} + static int is_swiotlb_buffer(phys_addr_t paddr) { return paddr >= virt_to_phys(io_tlb_start) && -- cgit v1.2.3 From ad32e8cb86e7894aac51c8963eaa9f36bb8a4e14 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Tue, 10 Nov 2009 19:46:19 +0900 Subject: swiotlb: Defer swiotlb init printing, export swiotlb_print_info() This enables us to avoid printing swiotlb memory info when we initialize swiotlb. After swiotlb initialization, we could find that we don't need swiotlb. This patch removes the code to print swiotlb memory info in swiotlb_init() and exports the function to do that. Signed-off-by: FUJITA Tomonori Cc: chrisw@sous-sol.org Cc: dwmw2@infradead.org Cc: joerg.roedel@amd.com Cc: muli@il.ibm.com Cc: tony.luck@intel.com Cc: benh@kernel.crashing.org LKML-Reference: <1257849980-22640-9-git-send-email-fujita.tomonori@lab.ntt.co.jp> [ -v2: merge up conflict ] Signed-off-by: Ingo Molnar --- arch/ia64/kernel/pci-swiotlb.c | 4 ++-- arch/powerpc/kernel/setup_32.c | 2 +- arch/powerpc/kernel/setup_64.c | 2 +- arch/x86/kernel/pci-swiotlb.c | 3 +-- include/linux/swiotlb.h | 4 ++-- lib/swiotlb.c | 15 ++++++++------- 6 files changed, 15 insertions(+), 15 deletions(-) (limited to 'lib') diff --git a/arch/ia64/kernel/pci-swiotlb.c b/arch/ia64/kernel/pci-swiotlb.c index 285aae8431c6..53292abf846c 100644 --- a/arch/ia64/kernel/pci-swiotlb.c +++ b/arch/ia64/kernel/pci-swiotlb.c @@ -41,7 +41,7 @@ struct dma_map_ops swiotlb_dma_ops = { void __init swiotlb_dma_init(void) { dma_ops = &swiotlb_dma_ops; - swiotlb_init(); + swiotlb_init(1); } void __init pci_swiotlb_init(void) @@ -51,7 +51,7 @@ void __init pci_swiotlb_init(void) swiotlb = 1; printk(KERN_INFO "PCI-DMA: Re-initialize machine vector.\n"); machvec_init("dig"); - swiotlb_init(); + swiotlb_init(1); dma_ops = &swiotlb_dma_ops; #else panic("Unable to find Intel IOMMU"); diff --git a/arch/powerpc/kernel/setup_32.c b/arch/powerpc/kernel/setup_32.c index 53bcf3d792db..b152de3e64d4 100644 --- a/arch/powerpc/kernel/setup_32.c +++ b/arch/powerpc/kernel/setup_32.c @@ -345,7 +345,7 @@ void __init setup_arch(char **cmdline_p) #ifdef CONFIG_SWIOTLB if (ppc_swiotlb_enable) - swiotlb_init(); + swiotlb_init(1); #endif paging_init(); diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c index 04f638d82fb3..df2c9e932b37 100644 --- a/arch/powerpc/kernel/setup_64.c +++ b/arch/powerpc/kernel/setup_64.c @@ -550,7 +550,7 @@ void __init setup_arch(char **cmdline_p) #ifdef CONFIG_SWIOTLB if (ppc_swiotlb_enable) - swiotlb_init(); + swiotlb_init(1); #endif paging_init(); diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c index aaa6b7839f1e..ea20ef7ca523 100644 --- a/arch/x86/kernel/pci-swiotlb.c +++ b/arch/x86/kernel/pci-swiotlb.c @@ -52,8 +52,7 @@ void __init pci_swiotlb_init(void) if (swiotlb_force) swiotlb = 1; if (swiotlb) { - printk(KERN_INFO "PCI-DMA: Using software bounce buffering for IO (SWIOTLB)\n"); - swiotlb_init(); + swiotlb_init(0); dma_ops = &swiotlb_dma_ops; } } diff --git a/include/linux/swiotlb.h b/include/linux/swiotlb.h index 59bafa690290..eb9bdb4d4854 100644 --- a/include/linux/swiotlb.h +++ b/include/linux/swiotlb.h @@ -20,8 +20,7 @@ struct scatterlist; */ #define IO_TLB_SHIFT 11 -extern void -swiotlb_init(void); +extern void swiotlb_init(int verbose); extern void *swiotlb_alloc_coherent(struct device *hwdev, size_t size, @@ -94,4 +93,5 @@ extern void __init swiotlb_free(void); static inline void swiotlb_free(void) { } #endif +extern void swiotlb_print_info(void); #endif /* __LINUX_SWIOTLB_H */ diff --git a/lib/swiotlb.c b/lib/swiotlb.c index eee512b63f17..0c12d7cce300 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -123,8 +123,9 @@ static dma_addr_t swiotlb_virt_to_bus(struct device *hwdev, return phys_to_dma(hwdev, virt_to_phys(address)); } -static void swiotlb_print_info(unsigned long bytes) +void swiotlb_print_info(void) { + unsigned long bytes = io_tlb_nslabs << IO_TLB_SHIFT; phys_addr_t pstart, pend; pstart = virt_to_phys(io_tlb_start); @@ -142,7 +143,7 @@ static void swiotlb_print_info(unsigned long bytes) * structures for the software IO TLB used to implement the DMA API. */ void __init -swiotlb_init_with_default_size(size_t default_size) +swiotlb_init_with_default_size(size_t default_size, int verbose) { unsigned long i, bytes; @@ -178,14 +179,14 @@ swiotlb_init_with_default_size(size_t default_size) io_tlb_overflow_buffer = alloc_bootmem_low(io_tlb_overflow); if (!io_tlb_overflow_buffer) panic("Cannot allocate SWIOTLB overflow buffer!\n"); - - swiotlb_print_info(bytes); + if (verbose) + swiotlb_print_info(); } void __init -swiotlb_init(void) +swiotlb_init(int verbose) { - swiotlb_init_with_default_size(64 * (1<<20)); /* default to 64MB */ + swiotlb_init_with_default_size(64 * (1<<20), verbose); /* default to 64MB */ } /* @@ -262,7 +263,7 @@ swiotlb_late_init_with_default_size(size_t default_size) if (!io_tlb_overflow_buffer) goto cleanup4; - swiotlb_print_info(bytes); + swiotlb_print_info(); late_alloc = 1; -- cgit v1.2.3 From 75f1cdf1dda92cae037ec848ae63690d91913eac Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Tue, 10 Nov 2009 19:46:20 +0900 Subject: x86: Handle HW IOMMU initialization failure gracefully If HW IOMMU initialization fails (Intel VT-d often does this, typically due to BIOS bugs), we fall back to nommu. It doesn't work for the majority since nowadays we have more than 4GB memory so we must use swiotlb instead of nommu. The problem is that it's too late to initialize swiotlb when HW IOMMU initialization fails. We need to allocate swiotlb memory earlier from bootmem allocator. Chris explained the issue in detail: http://marc.info/?l=linux-kernel&m=125657444317079&w=2 The current x86 IOMMU initialization sequence is too complicated and handling the above issue makes it more hacky. This patch changes x86 IOMMU initialization sequence to handle the above issue cleanly. The new x86 IOMMU initialization sequence are: 1. we initialize the swiotlb (and setting swiotlb to 1) in the case of (max_pfn > MAX_DMA32_PFN && !no_iommu). dma_ops is set to swiotlb_dma_ops or nommu_dma_ops. if swiotlb usage is forced by the boot option, we finish here. 2. we call the detection functions of all the IOMMUs 3. the detection function sets x86_init.iommu.iommu_init to the IOMMU initialization function (so we can avoid calling the initialization functions of all the IOMMUs needlessly). 4. if the IOMMU initialization function doesn't need to swiotlb then sets swiotlb to zero (e.g. the initialization is sucessful). 5. if we find that swiotlb is set to zero, we free swiotlb resource. Signed-off-by: FUJITA Tomonori Cc: chrisw@sous-sol.org Cc: dwmw2@infradead.org Cc: joerg.roedel@amd.com Cc: muli@il.ibm.com LKML-Reference: <1257849980-22640-10-git-send-email-fujita.tomonori@lab.ntt.co.jp> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/iommu.h | 1 - arch/x86/kernel/amd_iommu.c | 2 +- arch/x86/kernel/amd_iommu_init.c | 2 +- arch/x86/kernel/aperture_64.c | 2 +- arch/x86/kernel/pci-calgary_64.c | 10 +--------- arch/x86/kernel/pci-dma.c | 21 +++++++++++++-------- arch/x86/kernel/pci-gart_64.c | 1 + arch/x86/kernel/pci-nommu.c | 9 --------- arch/x86/kernel/pci-swiotlb.c | 7 +++---- drivers/pci/dmar.c | 3 +-- drivers/pci/intel-iommu.c | 6 ++++-- lib/swiotlb.c | 4 +++- 12 files changed, 29 insertions(+), 39 deletions(-) (limited to 'lib') diff --git a/arch/x86/include/asm/iommu.h b/arch/x86/include/asm/iommu.h index 878b30715766..df42a712361f 100644 --- a/arch/x86/include/asm/iommu.h +++ b/arch/x86/include/asm/iommu.h @@ -2,7 +2,6 @@ #define _ASM_X86_IOMMU_H static inline void iommu_shutdown_noop(void) {} -extern void no_iommu_init(void); extern struct dma_map_ops nommu_dma_ops; extern int force_iommu, no_iommu; extern int iommu_detected; diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 0285521e0a99..66237fde758f 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -2110,8 +2110,8 @@ int __init amd_iommu_init_dma_ops(void) prealloc_protection_domains(); iommu_detected = 1; - force_iommu = 1; bad_dma_address = 0; + swiotlb = 0; #ifdef CONFIG_GART_IOMMU gart_iommu_aperture_disabled = 1; gart_iommu_aperture = 0; diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index c41aabddaa2a..0d4581e602a4 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -1330,7 +1330,7 @@ static int __init early_amd_iommu_detect(struct acpi_table_header *table) void __init amd_iommu_detect(void) { - if (swiotlb || no_iommu || (iommu_detected && !gart_iommu_aperture)) + if (no_iommu || (iommu_detected && !gart_iommu_aperture)) return; if (acpi_table_parse("IVRS", early_amd_iommu_detect) == 0) { diff --git a/arch/x86/kernel/aperture_64.c b/arch/x86/kernel/aperture_64.c index 03933cf0b63c..e0dfb6856aa2 100644 --- a/arch/x86/kernel/aperture_64.c +++ b/arch/x86/kernel/aperture_64.c @@ -458,7 +458,7 @@ out: if (aper_alloc) { /* Got the aperture from the AGP bridge */ - } else if (swiotlb && !valid_agp) { + } else if (!valid_agp) { /* Do nothing */ } else if ((!no_iommu && max_pfn > MAX_DMA32_PFN) || force_iommu || diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 47bd419ea4d2..833f491440b9 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -1360,7 +1360,7 @@ void __init detect_calgary(void) * if the user specified iommu=off or iommu=soft or we found * another HW IOMMU already, bail out. */ - if (swiotlb || no_iommu || iommu_detected) + if (no_iommu || iommu_detected) return; if (!use_calgary) @@ -1445,10 +1445,6 @@ void __init detect_calgary(void) printk(KERN_INFO "PCI-DMA: Calgary TCE table spec is %d\n", specified_table_size); - /* swiotlb for devices that aren't behind the Calgary. */ - if (max_pfn > MAX_DMA32_PFN) - swiotlb = 1; - x86_init.iommu.iommu_init = calgary_iommu_init; } return; @@ -1476,11 +1472,7 @@ int __init calgary_iommu_init(void) return ret; } - force_iommu = 1; bad_dma_address = 0x0; - /* dma_ops is set to swiotlb or nommu */ - if (!dma_ops) - dma_ops = &nommu_dma_ops; return 0; } diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index bed05e2e5890..a234e63c2656 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -124,24 +124,24 @@ static void __init dma32_free_bootmem(void) void __init pci_iommu_alloc(void) { + /* swiotlb is forced by the boot option */ + int use_swiotlb = swiotlb; #ifdef CONFIG_X86_64 /* free the range so iommu could get some range less than 4G */ dma32_free_bootmem(); #endif + pci_swiotlb_init(); + if (use_swiotlb) + return; - /* - * The order of these functions is important for - * fall-back/fail-over reasons - */ gart_iommu_hole_init(); detect_calgary(); detect_intel_iommu(); + /* needs to be called after gart_iommu_hole_init */ amd_iommu_detect(); - - pci_swiotlb_init(); } void *dma_generic_alloc_coherent(struct device *dev, size_t size, @@ -291,10 +291,15 @@ static int __init pci_iommu_init(void) #ifdef CONFIG_PCI dma_debug_add_bus(&pci_bus_type); #endif - x86_init.iommu.iommu_init(); - no_iommu_init(); + if (swiotlb) { + printk(KERN_INFO "PCI-DMA: " + "Using software bounce buffering for IO (SWIOTLB)\n"); + swiotlb_print_info(); + } else + swiotlb_free(); + return 0; } /* Must execute after PCI subsystem */ diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index 0410bd30060d..919182e15d1e 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -833,6 +833,7 @@ int __init gart_iommu_init(void) flush_gart(); dma_ops = &gart_dma_ops; x86_platform.iommu_shutdown = gart_iommu_shutdown; + swiotlb = 0; return 0; } diff --git a/arch/x86/kernel/pci-nommu.c b/arch/x86/kernel/pci-nommu.c index a3933d4330cd..875e3822ae61 100644 --- a/arch/x86/kernel/pci-nommu.c +++ b/arch/x86/kernel/pci-nommu.c @@ -103,12 +103,3 @@ struct dma_map_ops nommu_dma_ops = { .sync_sg_for_device = nommu_sync_sg_for_device, .is_phys = 1, }; - -void __init no_iommu_init(void) -{ - if (dma_ops) - return; - - force_iommu = 0; /* no HW IOMMU */ - dma_ops = &nommu_dma_ops; -} diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c index ea20ef7ca523..17ce4221bd03 100644 --- a/arch/x86/kernel/pci-swiotlb.c +++ b/arch/x86/kernel/pci-swiotlb.c @@ -46,13 +46,12 @@ void __init pci_swiotlb_init(void) { /* don't initialize swiotlb if iommu=off (no_iommu=1) */ #ifdef CONFIG_X86_64 - if ((!iommu_detected && !no_iommu && max_pfn > MAX_DMA32_PFN)) + if (!no_iommu && max_pfn > MAX_DMA32_PFN) swiotlb = 1; #endif - if (swiotlb_force) - swiotlb = 1; if (swiotlb) { swiotlb_init(0); dma_ops = &swiotlb_dma_ops; - } + } else + dma_ops = &nommu_dma_ops; } diff --git a/drivers/pci/dmar.c b/drivers/pci/dmar.c index bce9cd7c755a..437399667e5a 100644 --- a/drivers/pci/dmar.c +++ b/drivers/pci/dmar.c @@ -613,8 +613,7 @@ void __init detect_intel_iommu(void) "x2apic and Intr-remapping.\n"); #endif #ifdef CONFIG_DMAR - if (ret && !no_iommu && !iommu_detected && !swiotlb && - !dmar_disabled) + if (ret && !no_iommu && !iommu_detected && !dmar_disabled) iommu_detected = 1; #endif #ifdef CONFIG_X86 diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c index b1e97e682500..43d755a2e14a 100644 --- a/drivers/pci/intel-iommu.c +++ b/drivers/pci/intel-iommu.c @@ -3231,7 +3231,7 @@ int __init intel_iommu_init(void) * Check the need for DMA-remapping initialization now. * Above initialization will also be used by Interrupt-remapping. */ - if (no_iommu || swiotlb || dmar_disabled) + if (no_iommu || dmar_disabled) return -ENODEV; iommu_init_mempool(); @@ -3252,7 +3252,9 @@ int __init intel_iommu_init(void) "PCI-DMA: Intel(R) Virtualization Technology for Directed I/O\n"); init_timer(&unmap_timer); - force_iommu = 1; +#ifdef CONFIG_SWIOTLB + swiotlb = 0; +#endif dma_ops = &intel_dma_ops; init_iommu_sysfs(); diff --git a/lib/swiotlb.c b/lib/swiotlb.c index 0c12d7cce300..e6755a0574fb 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -109,8 +109,10 @@ setup_io_tlb_npages(char *str) } if (*str == ',') ++str; - if (!strcmp(str, "force")) + if (!strcmp(str, "force")) { swiotlb_force = 1; + swiotlb = 1; + } return 1; } __setup("swiotlb=", setup_io_tlb_npages); -- cgit v1.2.3 From b18485e7acfe1a634615d1c628ef644c0d58d472 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Thu, 12 Nov 2009 00:03:28 +0900 Subject: swiotlb: Remove the swiotlb variable usage POWERPC doesn't expect it to be used. This fixes the linux-next build failure reported by Stephen Rothwell: lib/swiotlb.c: In function 'setup_io_tlb_npages': lib/swiotlb.c:114: error: 'swiotlb' undeclared (first use in this function) Reported-by: Stephen Rothwell Signed-off-by: FUJITA Tomonori Cc: peterz@infradead.org LKML-Reference: <20091112000258F.fujita.tomonori@lab.ntt.co.jp> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/swiotlb.h | 5 +++-- arch/x86/kernel/pci-dma.c | 5 +---- arch/x86/kernel/pci-swiotlb.c | 13 ++++++++++++- lib/swiotlb.c | 5 ++--- 4 files changed, 18 insertions(+), 10 deletions(-) (limited to 'lib') diff --git a/arch/x86/include/asm/swiotlb.h b/arch/x86/include/asm/swiotlb.h index b9e4e20174fb..940f13a213f8 100644 --- a/arch/x86/include/asm/swiotlb.h +++ b/arch/x86/include/asm/swiotlb.h @@ -9,11 +9,12 @@ extern int swiotlb_force; #ifdef CONFIG_SWIOTLB extern int swiotlb; -extern void pci_swiotlb_init(void); +extern int pci_swiotlb_init(void); #else #define swiotlb 0 -static inline void pci_swiotlb_init(void) +static inline int pci_swiotlb_init(void) { + return 0; } #endif diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index f79870e89266..0b11bf18f540 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -125,16 +125,13 @@ static void __init dma32_free_bootmem(void) void __init pci_iommu_alloc(void) { - /* swiotlb is forced by the boot option */ - int use_swiotlb = swiotlb; #ifdef CONFIG_X86_64 /* free the range so iommu could get some range less than 4G */ dma32_free_bootmem(); #else dma_ops = &nommu_dma_ops; #endif - pci_swiotlb_init(); - if (use_swiotlb) + if (pci_swiotlb_init()) return; gart_iommu_hole_init(); diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c index 17ce4221bd03..a6e5d0ffa3a7 100644 --- a/arch/x86/kernel/pci-swiotlb.c +++ b/arch/x86/kernel/pci-swiotlb.c @@ -42,16 +42,27 @@ static struct dma_map_ops swiotlb_dma_ops = { .dma_supported = NULL, }; -void __init pci_swiotlb_init(void) +/* + * pci_swiotlb_init - initialize swiotlb if necessary + * + * This returns non-zero if we are forced to use swiotlb (by the boot + * option). + */ +int __init pci_swiotlb_init(void) { /* don't initialize swiotlb if iommu=off (no_iommu=1) */ #ifdef CONFIG_X86_64 if (!no_iommu && max_pfn > MAX_DMA32_PFN) swiotlb = 1; #endif + if (swiotlb_force) + swiotlb = 1; + if (swiotlb) { swiotlb_init(0); dma_ops = &swiotlb_dma_ops; } else dma_ops = &nommu_dma_ops; + + return swiotlb_force; } diff --git a/lib/swiotlb.c b/lib/swiotlb.c index e6755a0574fb..795472d8ae24 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -109,10 +109,9 @@ setup_io_tlb_npages(char *str) } if (*str == ',') ++str; - if (!strcmp(str, "force")) { + if (!strcmp(str, "force")) swiotlb_force = 1; - swiotlb = 1; - } + return 1; } __setup("swiotlb=", setup_io_tlb_npages); -- cgit v1.2.3 From 8bfb2f8e655b9d0c45fde679fcd5fd97e34513db Mon Sep 17 00:00:00 2001 From: "Paul E. McKenney" Date: Wed, 2 Dec 2009 12:10:16 -0800 Subject: rcu: Make RCU's CPU-stall detector be default The RCU_CPU_STALL_DETECTOR costs almost nothing and has located some bugs that might otherwise have been difficult to track down. Make it be default for the TREE RCU implementations. The vmlinux size impact is limited (on 64-bit x86 defconfig): text data bss dec hex filename 8440248 1260076 995588 10695912 a334e8 vmlinux.before 8440774 1260060 995588 10696422 a336e6 vmlinux.after +526 bytes - acceptable default cost. For RAM starved systems, TINY_RCU does not support CPU-stall detection and is much smaller, but then again it is a uniprocessor... Signed-off-by: Paul E. McKenney Acked-by: Lai Jiangshan Cc: dipankar@in.ibm.com Cc: mathieu.desnoyers@polymtl.ca Cc: josh@joshtriplett.org Cc: dvhltc@us.ibm.com Cc: niv@us.ibm.com Cc: peterz@infradead.org Cc: rostedt@goodmis.org Cc: Valdis.Kletnieks@vt.edu Cc: dhowells@redhat.com LKML-Reference: <12597846162906-git-send-email-> [ v2: added image size calculations to the changelog ] Signed-off-by: Ingo Molnar --- lib/Kconfig.debug | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'lib') diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index 891155817bc6..50e0e78259e7 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -749,7 +749,7 @@ config RCU_TORTURE_TEST_RUNNABLE config RCU_CPU_STALL_DETECTOR bool "Check for stalled CPUs delaying RCU grace periods" depends on TREE_RCU || TREE_PREEMPT_RCU - default n + default y help This option causes RCU to printk information on which CPUs are delaying the current grace period, but only when -- cgit v1.2.3