From f5899b5d4fa806403f547dc41312d017d94ec273 Mon Sep 17 00:00:00 2001 From: John Hawkes Date: Fri, 16 Dec 2005 10:00:24 -0800 Subject: [IA64] disable preemption in udelay() The udelay() inline for ia64 uses the ITC. If CONFIG_PREEMPT is enabled and the platform has unsynchronized ITCs and the calling task migrates to another CPU while doing the udelay loop, then the effective delay may be too short or very, very long. This patch disables preemption around 100 usec chunks of the overall desired udelay time. This minimizes preemption-holdoffs. udelay() is now too big to be inline, move it out of line and export it. Signed-off-by: John Hawkes Signed-off-by: Tony Luck --- arch/ia64/kernel/time.c | 29 +++++++++++++++++++++++++++++ include/asm-ia64/delay.h | 10 +--------- 2 files changed, 30 insertions(+), 9 deletions(-) diff --git a/arch/ia64/kernel/time.c b/arch/ia64/kernel/time.c index 5b7e736f3b4..028a2b95936 100644 --- a/arch/ia64/kernel/time.c +++ b/arch/ia64/kernel/time.c @@ -249,3 +249,32 @@ time_init (void) */ set_normalized_timespec(&wall_to_monotonic, -xtime.tv_sec, -xtime.tv_nsec); } + +#define SMALLUSECS 100 + +void +udelay (unsigned long usecs) +{ + unsigned long start; + unsigned long cycles; + unsigned long smallusecs; + + /* + * Execute the non-preemptible delay loop (because the ITC might + * not be synchronized between CPUS) in relatively short time + * chunks, allowing preemption between the chunks. + */ + while (usecs > 0) { + smallusecs = (usecs > SMALLUSECS) ? SMALLUSECS : usecs; + preempt_disable(); + cycles = smallusecs*local_cpu_data->cyc_per_usec; + start = ia64_get_itc(); + + while (ia64_get_itc() - start < cycles) + cpu_relax(); + + preempt_enable(); + usecs -= smallusecs; + } +} +EXPORT_SYMBOL(udelay); diff --git a/include/asm-ia64/delay.h b/include/asm-ia64/delay.h index 57182d6f2b9..bba70207639 100644 --- a/include/asm-ia64/delay.h +++ b/include/asm-ia64/delay.h @@ -84,14 +84,6 @@ __delay (unsigned long loops) ia64_delay_loop (loops - 1); } -static __inline__ void -udelay (unsigned long usecs) -{ - unsigned long start = ia64_get_itc(); - unsigned long cycles = usecs*local_cpu_data->cyc_per_usec; - - while (ia64_get_itc() - start < cycles) - cpu_relax(); -} +extern void udelay (unsigned long usecs); #endif /* _ASM_IA64_DELAY_H */ -- cgit v1.2.3 From 3bd7f01713f30e7c616ab975ebb84ab7eb58a60a Mon Sep 17 00:00:00 2001 From: Jes Sorensen Date: Fri, 16 Dec 2005 11:00:03 -0500 Subject: [IA64] uncached ref count leak Use raw_smp_processor_id() instead of get_cpu() as we don't need the extra features of get_cpu(). Signed-off-by: Jes Sorensen Signed-off-by: Tony Luck --- arch/ia64/kernel/uncached.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/arch/ia64/kernel/uncached.c b/arch/ia64/kernel/uncached.c index c6d40446c2c..b631cf86ed4 100644 --- a/arch/ia64/kernel/uncached.c +++ b/arch/ia64/kernel/uncached.c @@ -53,7 +53,7 @@ static void uncached_ipi_visibility(void *data) if ((status != PAL_VISIBILITY_OK) && (status != PAL_VISIBILITY_OK_REMOTE_NEEDED)) printk(KERN_DEBUG "pal_prefetch_visibility() returns %i on " - "CPU %i\n", status, get_cpu()); + "CPU %i\n", status, raw_smp_processor_id()); } @@ -63,7 +63,7 @@ static void uncached_ipi_mc_drain(void *data) status = ia64_pal_mc_drain(); if (status) printk(KERN_WARNING "ia64_pal_mc_drain() failed with %i on " - "CPU %i\n", status, get_cpu()); + "CPU %i\n", status, raw_smp_processor_id()); } @@ -105,7 +105,7 @@ uncached_get_new_chunk(struct gen_pool *poolp) status = ia64_pal_prefetch_visibility(PAL_VISIBILITY_PHYSICAL); dprintk(KERN_INFO "pal_prefetch_visibility() returns %i on cpu %i\n", - status, get_cpu()); + status, raw_smp_processor_id()); if (!status) { status = smp_call_function(uncached_ipi_visibility, NULL, 0, 1); -- cgit v1.2.3 From d74700e604db717eef7a3112176e6350fb00d0e3 Mon Sep 17 00:00:00 2001 From: Jack Steiner Date: Thu, 15 Dec 2005 12:41:22 -0600 Subject: [IA64-SGI] Missed TLB flush I see why the problem exists only on SN. SN uses a different hardware mechanism to purge TLB entries across nodes. It looks like there is a bug in the SN TLB flushing code. During context switch, kernel threads inherit the mm of the task that was previously running on the cpu. This confuses the code in sn2_global_tlb_purge(). The result is a missed TLB purge for the task that owns the "borrowed" mm. (I hit the problem running heavy stress where kswapd was purging code pages of a user task that woke kswapd. The user task took a SIGILL fault trying to execute code in the page that had been ripped out from underneath it). Signed-off-by: Jack Steiner Signed-off-by: Tony Luck --- arch/ia64/sn/kernel/sn2/sn2_smp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/ia64/sn/kernel/sn2/sn2_smp.c b/arch/ia64/sn/kernel/sn2/sn2_smp.c index 5d54f5f4e92..471bbaa65d1 100644 --- a/arch/ia64/sn/kernel/sn2/sn2_smp.c +++ b/arch/ia64/sn/kernel/sn2/sn2_smp.c @@ -202,7 +202,7 @@ sn2_global_tlb_purge(struct mm_struct *mm, unsigned long start, unsigned long end, unsigned long nbits) { int i, opt, shub1, cnode, mynasid, cpu, lcpu = 0, nasid, flushed = 0; - int mymm = (mm == current->active_mm); + int mymm = (mm == current->active_mm && current->mm); volatile unsigned long *ptc0, *ptc1; unsigned long itc, itc2, flags, data0 = 0, data1 = 0, rr_value; short nasids[MAX_NUMNODES], nix; -- cgit v1.2.3 From d5bf3165b6fbb879a4658f9da9ca2fe002b75f08 Mon Sep 17 00:00:00 2001 From: "hawkes@sgi.com" Date: Tue, 13 Dec 2005 13:45:44 -0800 Subject: [IA64-SGI] change default_sn2 to NR_CPUS==1024 Change the NR_CPUS default for ia64/sn up to 1024. Signed-off-by: John Hawkes Signed-off-by: John Hesterberg Signed-off-by: Tony Luck --- arch/ia64/configs/sn2_defconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/ia64/configs/sn2_defconfig b/arch/ia64/configs/sn2_defconfig index e1924cc9687..ff8bb3770c9 100644 --- a/arch/ia64/configs/sn2_defconfig +++ b/arch/ia64/configs/sn2_defconfig @@ -113,7 +113,7 @@ CONFIG_IOSAPIC=y CONFIG_IA64_SGI_SN_XP=m CONFIG_FORCE_MAX_ZONEORDER=17 CONFIG_SMP=y -CONFIG_NR_CPUS=512 +CONFIG_NR_CPUS=1024 # CONFIG_HOTPLUG_CPU is not set CONFIG_SCHED_SMT=y CONFIG_PREEMPT=y -- cgit v1.2.3 From dc86e88c2bb8a7603ee175fbb6a9e92cf3293dd8 Mon Sep 17 00:00:00 2001 From: Christoph Lameter Date: Mon, 12 Dec 2005 09:34:32 -0800 Subject: [IA64] Add __read_mostly support for IA64 sparc64, i386 and x86_64 have support for a special data section dedicated to rarely updated data that is frequently read. The section was created to avoid false sharing of those rarely read data with frequently written kernel data. This patch creates such a data section for ia64 and will group rarely written data into this section. Signed-off-by: Christoph Lameter Signed-off-by: Tony Luck --- arch/ia64/kernel/vmlinux.lds.S | 3 +++ include/linux/cache.h | 2 +- 2 files changed, 4 insertions(+), 1 deletion(-) diff --git a/arch/ia64/kernel/vmlinux.lds.S b/arch/ia64/kernel/vmlinux.lds.S index 30d8564e960..73af6267d2e 100644 --- a/arch/ia64/kernel/vmlinux.lds.S +++ b/arch/ia64/kernel/vmlinux.lds.S @@ -177,6 +177,9 @@ SECTIONS } . = ALIGN(PAGE_SIZE); /* make sure the gate page doesn't expose kernel data */ + .data.read_mostly : AT(ADDR(.data.read_mostly) - LOAD_OFFSET) + { *(.data.read_mostly) } + .data.cacheline_aligned : AT(ADDR(.data.cacheline_aligned) - LOAD_OFFSET) { *(.data.cacheline_aligned) } diff --git a/include/linux/cache.h b/include/linux/cache.h index f6b5a46c5f8..0b7ecf3af78 100644 --- a/include/linux/cache.h +++ b/include/linux/cache.h @@ -13,7 +13,7 @@ #define SMP_CACHE_BYTES L1_CACHE_BYTES #endif -#if defined(CONFIG_X86) || defined(CONFIG_SPARC64) +#if defined(CONFIG_X86) || defined(CONFIG_SPARC64) || defined(CONFIG_IA64) #define __read_mostly __attribute__((__section__(".data.read_mostly"))) #else #define __read_mostly -- cgit v1.2.3