Random crashes when calling git

Random crashes when calling git

[Rolf Eike Beer]

[-- Attachment #1: Type: text/plain, Size: 3028 bytes --]

The CMake test suite includes running some git operations. I see the git test 
failing every some days, e.g. here:

http://open.cdash.org/testDetails.php?test=164070366&build=2734103

The syslog output to this event is this:

Dec 27 05:14:38 voyager kernel: 
Dec 27 05:14:38 voyager kernel: do_page_fault() pid=21120 command='git' type=6 
address=0x00000003
Dec 27 05:14:38 voyager kernel: 
Dec 27 05:14:38 voyager kernel: YZrvWESTHLNXBCVMcbcbcbcbOGFRQPDI
Dec 27 05:14:38 voyager kernel: PSW: 00000000000001000000000000001111 Not 
tainted
Dec 27 05:14:38 voyager kernel: r00-03  0004000f 4083bb40 00000000 00000007
Dec 27 05:14:38 voyager kernel: r04-07  4036e6f8 4103b480 00000000 4083c000
Dec 27 05:14:38 voyager kernel: r08-11  000000fb 4036b624 403709a4 00000000
Dec 27 05:14:38 voyager kernel: r12-15  402ffc60 fffff000 00800000 001397ae
Dec 27 05:14:38 voyager kernel: r16-19  fb217954 fb217964 4083c000 00000000
Dec 27 05:14:38 voyager kernel: r20-23  00000003 00000054 00000049 4103b900
Dec 27 05:14:38 voyager kernel: r24-27  0000000e 00123ec2 00000000 001350fc
Dec 27 05:14:38 voyager kernel: r28-31  00000000 00000002 4083c280 fefefeff
Dec 27 05:14:38 voyager kernel: sr00-03  0000002c 00000000 00000000 0000002c
Dec 27 05:14:38 voyager kernel: sr04-07  0000002c 0000002c 0000002c 0000002c
Dec 27 05:14:38 voyager kernel: 
Dec 27 05:14:38 voyager kernel: VZOUICununcqcqcqcqcqcrmunTDVZOUI
Dec 27 05:14:38 voyager kernel: FPSR: 00000000000000000000000000000000
Dec 27 05:14:38 voyager kernel: FPER1: 00000000
Dec 27 05:14:38 voyager kernel: fr00-03  0000000000000000 0000000000000000 
0000000000000000 0000000000000000
Dec 27 05:14:38 voyager kernel: fr04-07  0000000000000000 7d00000000000000 
00000004e2000000 4097700000000000
Dec 27 05:14:38 voyager kernel: fr08-11  0000000000000000 0000000000000000 
00000004019ff37b 2ecf5011101b4ce8
Dec 27 05:14:38 voyager kernel: fr12-15  4097700000000000 0000000000000000 
41d436ef72347f39 0000000050d30cd9
Dec 27 05:14:38 voyager kernel: fr16-19  0004fe0f105ad000 1012db6800000006 
0000000000000000 0000000000000031
Dec 27 05:14:38 voyager kernel: fr20-23  105ad00010605800 000000061060580a 
0000000000000088 00229d5b00000000
Dec 27 05:14:38 voyager kernel: fr24-27  0000000000000000 00000000a0b5ed8d 
00000000000000ff 0000000000000000
Dec 27 05:14:38 voyager kernel: fr28-31  00000005000203f9 105ad2b810592000 
105ad2a4000000ff 2f8302001011b974
Dec 27 05:14:38 voyager kernel: 
Dec 27 05:14:38 voyager kernel: IASQ: 0000002c 0000002c IAOQ: 00000003 
00000007
Dec 27 05:14:38 voyager kernel: IIR: 43ffff80    ISR: 0000002c  IOR: 4103b504
Dec 27 05:14:38 voyager kernel: CPU:        0   CR30: 2f100000 CR31: ffffffff
Dec 27 05:14:38 voyager kernel: ORIG_R28: 00000000
Dec 27 05:14:38 voyager kernel: IAOQ[0]: 00000003
Dec 27 05:14:38 voyager kernel: IAOQ[1]: 00000007
Dec 27 05:14:38 voyager kernel: RP(r2): 00000000

Kernel is up to date:

Linux voyager 3.7.1 #1 Thu Dec 20 12:22:22 CET 2012 parisc PA8600 (PCX-W+) 
9000/785/C3600 GNU/Linux

Glibc is 2.14.1.

Any ideas?

Eike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: Random crashes when calling git

[John David Anglin]

On 27-Dec-12, at 4:42 AM, Rolf Eike Beer wrote:

> Any ideas?

Try running the test under gdb.  Git has branched to location 0 causing
a page fault.  We need a backtrace(s) to learn something useful.  There
may be multiple backtraces (info thread).

Enable core dumps if not enabled.  Might find something useful by
examining it (gdb -c core /usr/bin/git).

Git uses threads.  I expect the problem is thread related since it  
doesn't
happen all the time, and for that reason it might not be reproducible  
running
under gdb..

You might try building git and running its testsuite.  I remember that
some thread related tests used to fail for me.  I think it may be  
possible
to build git without pthread support.

Dave
--
John David Anglin	dave.anglin@bell.net

Re: Random crashes when calling git

[Rolf Eike Beer]

[-- Attachment #1: Type: text/plain, Size: 892 bytes --]

John David Anglin wrote:
> On 27-Dec-12, at 4:42 AM, Rolf Eike Beer wrote:
> > Any ideas?
> 
> Try running the test under gdb.  Git has branched to location 0 causing
> a page fault.  We need a backtrace(s) to learn something useful.  There
> may be multiple backtraces (info thread).

I have not managed to reproduce this this way.

> Enable core dumps if not enabled.  Might find something useful by
> examining it (gdb -c core /usr/bin/git).
> 
> Git uses threads.  I expect the problem is thread related since it
> doesn't
> happen all the time, and for that reason it might not be reproducible
> running under gdb..
> 
> You might try building git and running its testsuite.  I remember that
> some thread related tests used to fail for me.  I think it may be
> possible to build git without pthread support.

I got 2 failures in 5400, 1 in 5406 and one in 5500 so far (in 4 runs).

Eike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: Random crashes when calling git

[Rolf Eike Beer]

[-- Attachment #1: Type: text/plain, Size: 905 bytes --]

John David Anglin wrote:
> On 27-Dec-12, at 4:42 AM, Rolf Eike Beer wrote:
> > Any ideas?
> 
> Try running the test under gdb.  Git has branched to location 0 causing
> a page fault.  We need a backtrace(s) to learn something useful.  There
> may be multiple backtraces (info thread).

This did not lead anywhere. I tried running the test only with strace, but 
even that did not lead to any crash.

> Enable core dumps if not enabled.  Might find something useful by
> examining it (gdb -c core /usr/bin/git).
> 
> Git uses threads.  I expect the problem is thread related since it
> doesn't
> happen all the time, and for that reason it might not be reproducible
> running under gdb..

I've put together the used git executable as well as 4 core dumps at 
http://opensource.sf-tec.de/git-crash.tar.bz2

It would be nice if you could take a look and tell me if this gives you 
anything.

Greetings,

Eike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: Random crashes when calling git

[John David Anglin]

[-- Attachment #1: Type: text/plain, Size: 3284 bytes --]

On 31-Dec-12, at 6:03 AM, Rolf Eike Beer wrote:

> I've put together the used git executable as well as 4 core dumps at
> http://opensource.sf-tec.de/git-crash.tar.bz2
>
> It would be nice if you could take a look and tell me if this gives  
> you
> anything.

I looked at the first one:

(gdb) p/x $pc
$1 = 0x0
(gdb) info threads
   Id   Target Id         Frame
   2    LWP 28212         0x000e054c in receive_status (in=1077339896,
     refs=0x19e578) at send-pack.c:163
* 1    LWP 28214         0x00000000 in ?? ()
(gdb) info reg
flags          0x4000f	262159
r1             0x40fe7b40	1090419520
rp             0x0	0
r3             0x7	7
r4             0x4036e6f8	1077339896
r5             0x417e7480	1098806400
r6             0x0	0
r7             0x40fe8000	1090420736
r8             0xfb	251
r9             0x4036b624	1077327396
r10            0x403709a4	1077348772
r11            0x0	0
r12            0x402ffc60	1076886624
r13            0xfffff000	4294963200
r14            0x800000	8388608
r15            0x14fcb2	1375410
r16            0xfb03d9d4	4211333588
r17            0xfb03d9e4	4211333604
r18            0x40fe8000	1090420736
r19            0x0	0
r20            0x3	3
r21            0x54	84
r22            0x49	73
---Type <return> to continue, or q <return> to quit---
r23            0x0	0
r24            0xe	14
r25            0x136bb2	1272754
r26            0x0	0
dp             0x14b0fc	1356028
ret0           0x0	0
ret1           0x2	2
sp             0x40fe8280	1090421376
r31            0xfefefeff	4278124287
sar            0x1d	29
pcoqh          0x3	3
pcsqh          0x2e	46
pcoqt          0x7	7
pcsqt          0x2e	46
eiem           0xfe000000	4261412864
iir            0x43ffff80	1140850560
isr            0x2e	46
ior            0x417e7504	1098806532
ipsw           0x4fe0f	327183
goto           *value not available*
sr4            0x2e	46
sr0            0x2e	46
sr1            0x0	0
(gdb) p/x $sp
$3 = 0x40fe8280
(gdb) x/64x $sp-64
0x40fe8240:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8250:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8260:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8270:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8280:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8290:	0x00007265	0x6d6f7465	0x3a013030	0x3065756e
0x40fe82a0:	0x7061636b	0x206f6b0a	0x30303139	0x6f6b2072
0x40fe82b0:	0x6566732f	0x68656164	0x732f6d61	0x73746572
0x40fe82c0:	0x0a303030	0x30000000	0x00000000	0x00000000
0x40fe82d0:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe82e0:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe82f0:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8300:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8310:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8320:	0x00000000	0x00000000	0x00000000	0x00000000
0x40fe8330:	0x00000000	0x00000000	0x00000000	0x00000000

I think this is the infamous thread stack corruption bug.  There are
numerous discussions about it on the list.

It's just a guess but I believe my kernel patch may help.  Attaching  
latest
version for 3.7.1.

If the patch doesn't help, then I think we need to review the glibc
patch sets for debian and gentoo.  I have installed an unmodified
version of 2.13-37.

Happy New Year,
Dave
--
John David Anglin	dave.anglin@bell.net


[-- Attachment #2: linux-stable-3.7.1-20121222.d.txt --]
[-- Type: text/plain, Size: 43292 bytes --]

diff --git a/arch/parisc/hpux/wrappers.S b/arch/parisc/hpux/wrappers.S
index 58c53c8..bdcea33 100644
--- a/arch/parisc/hpux/wrappers.S
+++ b/arch/parisc/hpux/wrappers.S
@@ -88,7 +88,7 @@ ENTRY(hpux_fork_wrapper)
 
 	STREG	%r2,-20(%r30)
 	ldo	64(%r30),%r30
-	STREG	%r2,PT_GR19(%r1)	;! save for child
+	STREG	%r2,PT_SYSCALL_RP(%r1)	;! save for child
 	STREG	%r30,PT_GR21(%r1)	;! save for child
 
 	LDREG	PT_GR30(%r1),%r25
@@ -132,7 +132,7 @@ ENTRY(hpux_child_return)
 	bl,n	schedule_tail, %r2
 #endif
 
-	LDREG	TASK_PT_GR19-TASK_SZ_ALGN-128(%r30),%r2
+	LDREG	TASK_PT_SYSCALL_RP-TASK_SZ_ALGN-128(%r30),%r2
 	b fork_return
 	copy %r0,%r28
 ENDPROC(hpux_child_return)
diff --git a/arch/parisc/include/asm/cacheflush.h b/arch/parisc/include/asm/cacheflush.h
index 9f21ab0..79f694f 100644
--- a/arch/parisc/include/asm/cacheflush.h
+++ b/arch/parisc/include/asm/cacheflush.h
@@ -115,7 +115,9 @@ flush_anon_page(struct vm_area_struct *vma, struct page *page, unsigned long vma
 {
 	if (PageAnon(page)) {
 		flush_tlb_page(vma, vmaddr);
+		preempt_disable();
 		flush_dcache_page_asm(page_to_phys(page), vmaddr);
+		preempt_enable();
 	}
 }
 
diff --git a/arch/parisc/include/asm/compat_rt_sigframe.h b/arch/parisc/include/asm/compat_rt_sigframe.h
deleted file mode 100644
index b3f95a7..0000000
--- a/arch/parisc/include/asm/compat_rt_sigframe.h
+++ /dev/null
@@ -1,50 +0,0 @@
-#include <linux/compat.h>
-#include <linux/compat_siginfo.h>
-#include <asm/compat_ucontext.h>
-
-#ifndef _ASM_PARISC_COMPAT_RT_SIGFRAME_H
-#define _ASM_PARISC_COMPAT_RT_SIGFRAME_H
-
-/* In a deft move of uber-hackery, we decide to carry the top half of all
- * 64-bit registers in a non-portable, non-ABI, hidden structure.
- * Userspace can read the hidden structure if it *wants* but is never
- * guaranteed to be in the same place. Infact the uc_sigmask from the 
- * ucontext_t structure may push the hidden register file downards
- */
-struct compat_regfile {
-	/* Upper half of all the 64-bit registers that were truncated
-	   on a copy to a 32-bit userspace */
-	compat_int_t rf_gr[32];
-	compat_int_t rf_iasq[2];
-	compat_int_t rf_iaoq[2];
-	compat_int_t rf_sar;
-};
-
-#define COMPAT_SIGRETURN_TRAMP 4
-#define COMPAT_SIGRESTARTBLOCK_TRAMP 5 
-#define COMPAT_TRAMP_SIZE (COMPAT_SIGRETURN_TRAMP + COMPAT_SIGRESTARTBLOCK_TRAMP)
-
-struct compat_rt_sigframe {
-	/* XXX: Must match trampoline size in arch/parisc/kernel/signal.c 
-	        Secondary to that it must protect the ERESTART_RESTARTBLOCK
-		trampoline we left on the stack (we were bad and didn't 
-		change sp so we could run really fast.) */
-	compat_uint_t tramp[COMPAT_TRAMP_SIZE];
-	compat_siginfo_t info;
-	struct compat_ucontext uc;
-	/* Hidden location of truncated registers, *must* be last. */
-	struct compat_regfile regs; 
-};
-
-/*
- * The 32-bit ABI wants at least 48 bytes for a function call frame:
- * 16 bytes for arg0-arg3, and 32 bytes for magic (the only part of
- * which Linux/parisc uses is sp-20 for the saved return pointer...)
- * Then, the stack pointer must be rounded to a cache line (64 bytes).
- */
-#define SIGFRAME32		64
-#define FUNCTIONCALLFRAME32	48
-#define PARISC_RT_SIGFRAME_SIZE32					\
-	(((sizeof(struct compat_rt_sigframe) + FUNCTIONCALLFRAME32) + SIGFRAME32) & -SIGFRAME32)
-
-#endif
diff --git a/arch/parisc/include/asm/mmzone.h b/arch/parisc/include/asm/mmzone.h
index e67eb9c..31835b9 100644
--- a/arch/parisc/include/asm/mmzone.h
+++ b/arch/parisc/include/asm/mmzone.h
@@ -1,9 +1,10 @@
 #ifndef _PARISC_MMZONE_H
 #define _PARISC_MMZONE_H
 
+#define MAX_PHYSMEM_RANGES 8 /* Fix the size for now (current known max is 3) */
+
 #ifdef CONFIG_DISCONTIGMEM
 
-#define MAX_PHYSMEM_RANGES 8 /* Fix the size for now (current known max is 3) */
 extern int npmem_ranges;
 
 struct node_map_data {
@@ -60,7 +61,5 @@ static inline int pfn_valid(int pfn)
 	return 0;
 }
 
-#else /* !CONFIG_DISCONTIGMEM */
-#define MAX_PHYSMEM_RANGES 	1 
 #endif
 #endif /* _PARISC_MMZONE_H */
diff --git a/arch/parisc/include/asm/page.h b/arch/parisc/include/asm/page.h
index 4e0e7db..d9812d8 100644
--- a/arch/parisc/include/asm/page.h
+++ b/arch/parisc/include/asm/page.h
@@ -21,15 +21,27 @@
 #include <asm/types.h>
 #include <asm/cache.h>
 
-#define clear_page(page)	memset((void *)(page), 0, PAGE_SIZE)
-#define copy_page(to,from)      copy_user_page_asm((void *)(to), (void *)(from))
+#define clear_page(page)	clear_page_asm((void *)(page))
+#define copy_page(to,from)	copy_page_asm((void *)(to), (void *)(from))
 
 struct page;
 
-void copy_user_page_asm(void *to, void *from);
+void clear_page_asm(void *page);
+void copy_page_asm(void *to, void *from);
+void clear_user_page(void *vto, unsigned long vaddr, struct page *pg);
 void copy_user_page(void *vto, void *vfrom, unsigned long vaddr,
 			   struct page *pg);
-void clear_user_page(void *page, unsigned long vaddr, struct page *pg);
+
+// #define CONFIG_PARISC_TMPALIAS
+
+#ifdef CONFIG_PARISC_TMPALIAS
+void clear_user_highpage(struct page *page, unsigned long vaddr);
+#define clear_user_highpage clear_user_highpage
+struct vm_area_struct;
+void copy_user_highpage(struct page *to, struct page *from,
+	unsigned long vaddr, struct vm_area_struct *vma);
+#define __HAVE_ARCH_COPY_USER_HIGHPAGE
+#endif
 
 /*
  * These are used to make use of C type-checking..
diff --git a/arch/parisc/include/asm/pgtable.h b/arch/parisc/include/asm/pgtable.h
index ee99f23..563724d 100644
--- a/arch/parisc/include/asm/pgtable.h
+++ b/arch/parisc/include/asm/pgtable.h
@@ -12,11 +12,10 @@
 
 #include <linux/bitops.h>
 #include <linux/spinlock.h>
+#include <linux/mm_types.h>
 #include <asm/processor.h>
 #include <asm/cache.h>
 
-struct vm_area_struct;
-
 /*
  * kern_addr_valid(ADDR) tests if ADDR is pointing to valid kernel
  * memory.  For the return value to be meaningful, ADDR must be >=
@@ -40,7 +39,14 @@ struct vm_area_struct;
         do{                                                     \
                 *(pteptr) = (pteval);                           \
         } while(0)
-#define set_pte_at(mm,addr,ptep,pteval) set_pte(ptep,pteval)
+
+extern void purge_tlb_entries(struct mm_struct *, unsigned long);
+
+#define set_pte_at(mm,addr,ptep, pteval)                        \
+        do{                                                     \
+                set_pte(ptep,pteval);                           \
+                purge_tlb_entries(mm,addr);                     \
+        } while(0)
 
 #endif /* !__ASSEMBLY__ */
 
@@ -462,10 +468,13 @@ static inline void ptep_set_wrprotect(struct mm_struct *mm, unsigned long addr,
 #ifdef CONFIG_SMP
 	unsigned long new, old;
 
+	/* ??? This might be racy because the page table updates in
+	   entry.S don't use the same lock.  */
 	do {
 		old = pte_val(*ptep);
 		new = pte_val(pte_wrprotect(__pte (old)));
 	} while (cmpxchg((unsigned long *) ptep, old, new) != old);
+	purge_tlb_entries(mm, addr);
 #else
 	pte_t old_pte = *ptep;
 	set_pte_at(mm, addr, ptep, pte_wrprotect(old_pte));
diff --git a/arch/parisc/kernel/asm-offsets.c b/arch/parisc/kernel/asm-offsets.c
index dcd5510..5df1597 100644
--- a/arch/parisc/kernel/asm-offsets.c
+++ b/arch/parisc/kernel/asm-offsets.c
@@ -141,6 +141,7 @@ int main(void)
 	DEFINE(TASK_PT_IAOQ0, offsetof(struct task_struct, thread.regs.iaoq[0]));
 	DEFINE(TASK_PT_IAOQ1, offsetof(struct task_struct, thread.regs.iaoq[1]));
 	DEFINE(TASK_PT_CR27, offsetof(struct task_struct, thread.regs.cr27));
+	DEFINE(TASK_PT_SYSCALL_RP, offsetof(struct task_struct, thread.regs.pad0));
 	DEFINE(TASK_PT_ORIG_R28, offsetof(struct task_struct, thread.regs.orig_r28));
 	DEFINE(TASK_PT_KSP, offsetof(struct task_struct, thread.regs.ksp));
 	DEFINE(TASK_PT_KPC, offsetof(struct task_struct, thread.regs.kpc));
@@ -230,6 +231,7 @@ int main(void)
 	DEFINE(PT_IAOQ0, offsetof(struct pt_regs, iaoq[0]));
 	DEFINE(PT_IAOQ1, offsetof(struct pt_regs, iaoq[1]));
 	DEFINE(PT_CR27, offsetof(struct pt_regs, cr27));
+	DEFINE(PT_SYSCALL_RP, offsetof(struct pt_regs, pad0));
 	DEFINE(PT_ORIG_R28, offsetof(struct pt_regs, orig_r28));
 	DEFINE(PT_KSP, offsetof(struct pt_regs, ksp));
 	DEFINE(PT_KPC, offsetof(struct pt_regs, kpc));
diff --git a/arch/parisc/kernel/cache.c b/arch/parisc/kernel/cache.c
index 48e16dc..fb449c3 100644
--- a/arch/parisc/kernel/cache.c
+++ b/arch/parisc/kernel/cache.c
@@ -267,9 +267,11 @@ static inline void
 __flush_cache_page(struct vm_area_struct *vma, unsigned long vmaddr,
 		   unsigned long physaddr)
 {
+	preempt_disable();
 	flush_dcache_page_asm(physaddr, vmaddr);
 	if (vma->vm_flags & VM_EXEC)
 		flush_icache_page_asm(physaddr, vmaddr);
+	preempt_enable();
 }
 
 void flush_dcache_page(struct page *page)
@@ -314,7 +316,7 @@ void flush_dcache_page(struct page *page)
 		flush_tlb_page(mpnt, addr);
 		if (old_addr == 0 || (old_addr & (SHMLBA - 1)) != (addr & (SHMLBA - 1))) {
 			__flush_cache_page(mpnt, addr, page_to_phys(page));
-			if (old_addr)
+			if (old_addr && parisc_requires_coherency())
 				printk(KERN_ERR "INEQUIVALENT ALIASES 0x%lx and 0x%lx in file %s\n", old_addr, addr, mpnt->vm_file ? (char *)mpnt->vm_file->f_path.dentry->d_name.name : "(null)");
 			old_addr = addr;
 		}
@@ -329,17 +331,6 @@ EXPORT_SYMBOL(flush_kernel_dcache_page_asm);
 EXPORT_SYMBOL(flush_data_cache_local);
 EXPORT_SYMBOL(flush_kernel_icache_range_asm);
 
-void clear_user_page_asm(void *page, unsigned long vaddr)
-{
-	unsigned long flags;
-	/* This function is implemented in assembly in pacache.S */
-	extern void __clear_user_page_asm(void *page, unsigned long vaddr);
-
-	purge_tlb_start(flags);
-	__clear_user_page_asm(page, vaddr);
-	purge_tlb_end(flags);
-}
-
 #define FLUSH_THRESHOLD 0x80000 /* 0.5MB */
 int parisc_cache_flush_threshold __read_mostly = FLUSH_THRESHOLD;
 
@@ -373,20 +364,9 @@ void __init parisc_setup_cache_timing(void)
 	printk(KERN_INFO "Setting cache flush threshold to %x (%d CPUs online)\n", parisc_cache_flush_threshold, num_online_cpus());
 }
 
-extern void purge_kernel_dcache_page(unsigned long);
-extern void clear_user_page_asm(void *page, unsigned long vaddr);
-
-void clear_user_page(void *page, unsigned long vaddr, struct page *pg)
-{
-	unsigned long flags;
-
-	purge_kernel_dcache_page((unsigned long)page);
-	purge_tlb_start(flags);
-	pdtlb_kernel(page);
-	purge_tlb_end(flags);
-	clear_user_page_asm(page, vaddr);
-}
-EXPORT_SYMBOL(clear_user_page);
+extern void purge_kernel_dcache_page_asm(unsigned long);
+extern void clear_user_page_asm(void *, unsigned long);
+extern void copy_user_page_asm(void *, void *, unsigned long);
 
 void flush_kernel_dcache_page_addr(void *addr)
 {
@@ -399,11 +379,26 @@ void flush_kernel_dcache_page_addr(void *addr)
 }
 EXPORT_SYMBOL(flush_kernel_dcache_page_addr);
 
+void clear_user_page(void *vto, unsigned long vaddr, struct page *page)
+{
+	clear_page_asm(vto);
+	if (!parisc_requires_coherency())
+		flush_kernel_dcache_page_asm(vto);
+}
+EXPORT_SYMBOL(clear_user_page);
+
 void copy_user_page(void *vto, void *vfrom, unsigned long vaddr,
-		    struct page *pg)
+	struct page *pg)
 {
-	/* no coherency needed (all in kmap/kunmap) */
-	copy_user_page_asm(vto, vfrom);
+	/* Copy using kernel mapping.  No coherency is needed
+	   (all in kmap/kunmap) on machines that don't support
+	   non-equivalent aliasing.  However, the `from' page
+	   needs to be flushed before it can be accessed through
+	   the kernel mapping. */
+	preempt_disable();
+	flush_dcache_page_asm(__pa(vfrom), vaddr);
+	preempt_enable();
+	copy_page_asm(vto, vfrom);
 	if (!parisc_requires_coherency())
 		flush_kernel_dcache_page_asm(vto);
 }
@@ -419,6 +414,25 @@ void kunmap_parisc(void *addr)
 EXPORT_SYMBOL(kunmap_parisc);
 #endif
 
+void purge_tlb_entries(struct mm_struct *mm, unsigned long addr)
+{
+	unsigned long flags, sid;
+
+	/* Note: purge_tlb_entries can be called at startup with
+	   no context.  */
+
+	/* Disable preemption while we play with %sr1.  */
+	preempt_disable();
+	sid = mfsp(1);
+	mtsp(mm->context,1);
+	purge_tlb_start(flags);
+	pdtlb(addr);
+	pitlb(addr);
+	purge_tlb_end(flags);
+	mtsp(sid,1);
+	preempt_enable();
+}
+
 void __flush_tlb_range(unsigned long sid, unsigned long start,
 		       unsigned long end)
 {
@@ -458,8 +472,65 @@ void flush_cache_all(void)
 	on_each_cpu(cacheflush_h_tmp_function, NULL, 1);
 }
 
+static inline unsigned long mm_total_size(struct mm_struct *mm)
+{
+	struct vm_area_struct *vma;
+	unsigned long usize = 0;
+
+	for (vma = mm->mmap; vma; vma = vma->vm_next)
+		usize += vma->vm_end - vma->vm_start;
+	return usize;
+}
+
+static inline pte_t *get_ptep(pgd_t *pgd, unsigned long addr)
+{
+	pte_t *ptep = NULL;
+
+        if (!pgd_none(*pgd)) {
+                pud_t *pud = pud_offset(pgd, addr);
+                if (!pud_none(*pud)) {
+                        pmd_t *pmd = pmd_offset(pud, addr);
+                        if (!pmd_none(*pmd)) {
+                                ptep = pte_offset_map(pmd, addr);
+                        }
+                }
+        }
+	return ptep;
+}
+
 void flush_cache_mm(struct mm_struct *mm)
 {
+	/* Flushing the whole cache on each cpu takes forever on
+	   rp3440, etc.  So, avoid it if the mm isn't too big.
+	   Note: This approach is faster than a range flush when the
+	   context is current, and it works even when non current.  */
+	if (mm_total_size(mm) < parisc_cache_flush_threshold) {
+		struct vm_area_struct *vma;
+
+		if (mm->context == mfsp(3)) {
+			for (vma = mm->mmap; vma; vma = vma->vm_next) {
+				flush_user_dcache_range_asm(vma->vm_start, vma->vm_end);
+				if(vma->vm_flags & VM_EXEC)
+					flush_user_icache_range_asm(vma->vm_start, vma->vm_end);
+			}
+		} else {
+			pgd_t *pgd = mm->pgd;
+
+			for (vma = mm->mmap; vma; vma = vma->vm_next) {
+				unsigned long addr;
+
+				for (addr = vma->vm_start; addr < vma->vm_end; addr += PAGE_SIZE) {
+					pte_t *ptep = get_ptep(pgd, addr);
+					if (ptep != NULL) {
+						pte_t pte = *ptep;
+						__flush_cache_page(vma, addr, page_to_phys(pte_page(pte)));
+					}
+				}
+			}
+		}
+		return;
+	}
+
 #ifdef CONFIG_SMP
 	flush_cache_all();
 #else
@@ -485,20 +556,34 @@ flush_user_icache_range(unsigned long start, unsigned long end)
 		flush_instruction_cache();
 }
 
-
 void flush_cache_range(struct vm_area_struct *vma,
 		unsigned long start, unsigned long end)
 {
-	int sr3;
-
 	BUG_ON(!vma->vm_mm->context);
 
-	sr3 = mfsp(3);
-	if (vma->vm_mm->context == sr3) {
-		flush_user_dcache_range(start,end);
-		flush_user_icache_range(start,end);
+	if ((end - start) < parisc_cache_flush_threshold) {
+		if (vma->vm_mm->context == mfsp(3)) {
+			flush_user_dcache_range_asm(start,end);
+			if(vma->vm_flags & VM_EXEC)
+				flush_user_icache_range_asm(start,end);
+		} else {
+			unsigned long addr;
+			pgd_t *pgd = vma->vm_mm->pgd;
+
+			for (addr = start & PAGE_MASK; addr < end; addr += PAGE_SIZE) {
+				pte_t *ptep = get_ptep(pgd, addr);
+				if (ptep != NULL) {
+					pte_t pte = *ptep;
+					flush_cache_page(vma, addr, pte_pfn(pte));
+				}
+			}
+		}
 	} else {
+#ifdef CONFIG_SMP
 		flush_cache_all();
+#else
+		flush_cache_all_local();
+#endif
 	}
 }
 
@@ -511,3 +596,67 @@ flush_cache_page(struct vm_area_struct *vma, unsigned long vmaddr, unsigned long
 	__flush_cache_page(vma, vmaddr, page_to_phys(pfn_to_page(pfn)));
 
 }
+
+#ifdef CONFIG_PARISC_TMPALIAS
+
+void clear_user_highpage(struct page *page, unsigned long vaddr)
+{
+	void *vto;
+	unsigned long flags;
+
+	/* Clear using TMPALIAS region.  The page doesn't need to
+	   be flushed but the kernel mapping needs to be purged.  */
+
+	vto = kmap_atomic(page, KM_USER0);
+
+	/* The PA-RISC 2.0 Architecture book states on page F-6:
+	   "Before a write-capable translation is enabled, *all*
+	   non-equivalently-aliased translations must be removed
+	   from the page table and purged from the TLB.  (Note
+	   that the caches are not required to be flushed at this
+	   time.)  Before any non-equivalent aliased translation
+	   is re-enabled, the virtual address range for the writeable
+	   page (the entire page) must be flushed from the cache,
+	   and the write-capable translation removed from the page
+	   table and purged from the TLB."  */
+
+	purge_kernel_dcache_page_asm((unsigned long)vto);
+	purge_tlb_start(flags);
+	pdtlb_kernel(vto);
+	purge_tlb_end(flags);
+	preempt_disable();
+	clear_user_page_asm(vto, vaddr);
+	preempt_enable();
+
+	pagefault_enable();		/* kunmap_atomic(addr, KM_USER0); */
+}
+
+void copy_user_highpage(struct page *to, struct page *from,
+	unsigned long vaddr, struct vm_area_struct *vma)
+{
+	void *vfrom, *vto;
+	unsigned long flags;
+
+	/* Copy using TMPALIAS region.  This has the advantage
+	   that the `from' page doesn't need to be flushed.  However,
+	   the `to' page must be flushed in copy_user_page_asm since
+	   it can be used to bring in executable code.  */
+
+	vfrom = kmap_atomic(from, KM_USER0);
+	vto = kmap_atomic(to, KM_USER1);
+
+	purge_kernel_dcache_page_asm((unsigned long)vto);
+	purge_tlb_start(flags);
+	pdtlb_kernel(vto);
+	pdtlb_kernel(vfrom);
+	purge_tlb_end(flags);
+	preempt_disable();
+	copy_user_page_asm(vto, vfrom, vaddr);
+	flush_dcache_page_asm(__pa(vto), vaddr);
+	preempt_enable();
+
+	pagefault_enable();		/* kunmap_atomic(addr, KM_USER1); */
+	pagefault_enable();		/* kunmap_atomic(addr, KM_USER0); */
+}
+
+#endif /* CONFIG_PARISC_TMPALIAS */
diff --git a/arch/parisc/kernel/entry.S b/arch/parisc/kernel/entry.S
index 18670a0..c19b2b0 100644
--- a/arch/parisc/kernel/entry.S
+++ b/arch/parisc/kernel/entry.S
@@ -483,7 +483,7 @@
 	 * B <-> _PAGE_DMB (memory break)
 	 *
 	 * Then incredible subtlety: The access rights are
-	 * _PAGE_GATEWAY _PAGE_EXEC _PAGE_READ
+	 * _PAGE_GATEWAY, _PAGE_EXEC and _PAGE_WRITE
 	 * See 3-14 of the parisc 2.0 manual
 	 *
 	 * Finally, _PAGE_READ goes in the top bit of PL1 (so we
@@ -493,7 +493,7 @@
 
 	/* PAGE_USER indicates the page can be read with user privileges,
 	 * so deposit X1|11 to PL1|PL2 (remember the upper bit of PL1
-	 * contains _PAGE_READ */
+	 * contains _PAGE_READ) */
 	extrd,u,*=      \pte,_PAGE_USER_BIT+32,1,%r0
 	depdi		7,11,3,\prot
 	/* If we're a gateway page, drop PL2 back to zero for promotion
@@ -1785,9 +1785,9 @@ ENTRY(sys_fork_wrapper)
 	ldo	-16(%r30),%r29		/* Reference param save area */
 #endif
 
-	/* These are call-clobbered registers and therefore
-	   also syscall-clobbered (we hope). */
-	STREG	%r2,PT_GR19(%r1)	/* save for child */
+	STREG	%r2,PT_SYSCALL_RP(%r1)	/* save for child */
+
+	/* WARNING - Clobbers r21, userspace must save! */
 	STREG	%r30,PT_GR21(%r1)
 
 	LDREG	PT_GR30(%r1),%r25
@@ -1817,7 +1817,7 @@ ENTRY(child_return)
 	nop
 
 	LDREG	TI_TASK-THREAD_SZ_ALGN-FRAME_SIZE-FRAME_SIZE(%r30), %r1
-	LDREG	TASK_PT_GR19(%r1),%r2
+	LDREG	TASK_PT_SYSCALL_RP(%r1),%r2
 	b	wrapper_exit
 	copy	%r0,%r28
 ENDPROC(child_return)
@@ -1836,8 +1836,9 @@ ENTRY(sys_clone_wrapper)
 	ldo	-16(%r30),%r29		/* Reference param save area */
 #endif
 
-	/* WARNING - Clobbers r19 and r21, userspace must save these! */
-	STREG	%r2,PT_GR19(%r1)	/* save for child */
+	STREG	%r2,PT_SYSCALL_RP(%r1)	/* save for child */
+
+	/* WARNING - Clobbers r21, userspace must save! */
 	STREG	%r30,PT_GR21(%r1)
 	BL	sys_clone,%r2
 	copy	%r1,%r24
@@ -1860,7 +1861,7 @@ ENTRY(sys_vfork_wrapper)
 	ldo	-16(%r30),%r29		/* Reference param save area */
 #endif
 
-	STREG	%r2,PT_GR19(%r1)	/* save for child */
+	STREG	%r2,PT_SYSCALL_RP(%r1)	/* save for child */
 	STREG	%r30,PT_GR21(%r1)
 
 	BL	sys_vfork,%r2
@@ -1893,15 +1894,7 @@ ENDPROC(sys_vfork_wrapper)
 
 	ldo -FRAME_SIZE(%r30),%r30
 	LDREG -RP_OFFSET(%r30),%r2
-
-	/* If exec succeeded we need to load the args */
-
-	ldo -1024(%r0),%r1
-	cmpb,>>= %r28,%r1,error_\execve
-	copy %r2,%r19
-
-error_\execve:
-	bv %r0(%r19)
+	bv,n %r0(%r2)
 	nop
 	.endm
 
@@ -2064,7 +2057,7 @@ syscall_restore:
 
 	/* Are we being ptraced? */
 	ldw	TASK_FLAGS(%r1),%r19
-	ldi	(_TIF_SINGLESTEP|_TIF_BLOCKSTEP),%r2
+	ldi	_TIF_SYSCALL_TRACE_MASK,%r2
 	and,COND(=)	%r19,%r2,%r0
 	b,n	syscall_restore_rfi
 
@@ -2177,15 +2170,23 @@ syscall_restore_rfi:
 	/* sr2 should be set to zero for userspace syscalls */
 	STREG	%r0,TASK_PT_SR2(%r1)
 
-pt_regs_ok:
 	LDREG	TASK_PT_GR31(%r1),%r2
-	depi	3,31,2,%r2			   /* ensure return to user mode. */
-	STREG	%r2,TASK_PT_IAOQ0(%r1)
+	depi	3,31,2,%r2		   /* ensure return to user mode. */
+	STREG   %r2,TASK_PT_IAOQ0(%r1)
 	ldo	4(%r2),%r2
 	STREG	%r2,TASK_PT_IAOQ1(%r1)
+	b	intr_restore
 	copy	%r25,%r16
+
+pt_regs_ok:
+	LDREG	TASK_PT_IAOQ0(%r1),%r2
+	depi	3,31,2,%r2		   /* ensure return to user mode. */
+	STREG	%r2,TASK_PT_IAOQ0(%r1)
+	LDREG	TASK_PT_IAOQ1(%r1),%r2
+	depi	3,31,2,%r2
+	STREG	%r2,TASK_PT_IAOQ1(%r1)
 	b	intr_restore
-	nop
+	copy	%r25,%r16
 
 	.import schedule,code
 syscall_do_resched:
diff --git a/arch/parisc/kernel/inventory.c b/arch/parisc/kernel/inventory.c
index 08324aa..3295ef4 100644
--- a/arch/parisc/kernel/inventory.c
+++ b/arch/parisc/kernel/inventory.c
@@ -186,12 +186,14 @@ pat_query_module(ulong pcell_loc, ulong mod_index)
 
 	if (status != PDC_OK) {
 		/* no more cell modules or error */
+		kfree(pa_pdc_cell);
 		return status;
 	}
 
 	temp = pa_pdc_cell->cba;
 	dev = alloc_pa_dev(PAT_GET_CBA(temp), &(pa_pdc_cell->mod_path));
 	if (!dev) {
+		kfree(pa_pdc_cell);
 		return PDC_OK;
 	}
 
diff --git a/arch/parisc/kernel/irq.c b/arch/parisc/kernel/irq.c
index c0b1aff..8094d3e 100644
--- a/arch/parisc/kernel/irq.c
+++ b/arch/parisc/kernel/irq.c
@@ -379,14 +379,14 @@ void do_cpu_irq_mask(struct pt_regs *regs)
 static struct irqaction timer_action = {
 	.handler = timer_interrupt,
 	.name = "timer",
-	.flags = IRQF_DISABLED | IRQF_TIMER | IRQF_PERCPU | IRQF_IRQPOLL,
+	.flags = IRQF_TIMER | IRQF_PERCPU | IRQF_IRQPOLL,
 };
 
 #ifdef CONFIG_SMP
 static struct irqaction ipi_action = {
 	.handler = ipi_interrupt,
 	.name = "IPI",
-	.flags = IRQF_DISABLED | IRQF_PERCPU,
+	.flags = IRQF_PERCPU,
 };
 #endif
 
@@ -410,11 +410,13 @@ void __init init_IRQ(void)
 {
 	local_irq_disable();	/* PARANOID - should already be disabled */
 	mtctl(~0UL, 23);	/* EIRR : clear all pending external intr */
-	claim_cpu_irqs();
 #ifdef CONFIG_SMP
-	if (!cpu_eiem)
+	if (!cpu_eiem) {
+		claim_cpu_irqs();
 		cpu_eiem = EIEM_MASK(IPI_IRQ) | EIEM_MASK(TIMER_IRQ);
+	}
 #else
+	claim_cpu_irqs();
 	cpu_eiem = EIEM_MASK(TIMER_IRQ);
 #endif
         set_eiem(cpu_eiem);	/* EIEM : enable all external intr */
diff --git a/arch/parisc/kernel/pacache.S b/arch/parisc/kernel/pacache.S
index 5d7218a..ed401dd 100644
--- a/arch/parisc/kernel/pacache.S
+++ b/arch/parisc/kernel/pacache.S
@@ -199,7 +199,6 @@ ENTRY(flush_instruction_cache_local)
 	.callinfo NO_CALLS
 	.entry
 
-	mtsp		%r0, %sr1
 	load32		cache_info, %r1
 
 	/* Flush Instruction Cache */
@@ -208,20 +207,46 @@ ENTRY(flush_instruction_cache_local)
 	LDREG		ICACHE_STRIDE(%r1), %arg1
 	LDREG		ICACHE_COUNT(%r1), %arg2
 	LDREG		ICACHE_LOOP(%r1), %arg3
-	rsm             PSW_SM_I, %r22		/* No mmgt ops during loop*/
+	rsm		PSW_SM_I, %r22		/* No mmgt ops during loop*/
 	addib,COND(=)		-1, %arg3, fioneloop	/* Preadjust and test */
 	movb,<,n	%arg3, %r31, fisync	/* If loop < 0, do sync */
 
 fimanyloop:					/* Loop if LOOP >= 2 */
 	addib,COND(>)		-1, %r31, fimanyloop	/* Adjusted inner loop decr */
-	fice            %r0(%sr1, %arg0)
-	fice,m		%arg1(%sr1, %arg0)	/* Last fice and addr adjust */
+	fice            %r0(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)	/* Last fice and addr adjust */
 	movb,tr		%arg3, %r31, fimanyloop	/* Re-init inner loop count */
 	addib,COND(<=),n	-1, %arg2, fisync	/* Outer loop decr */
 
 fioneloop:					/* Loop if LOOP = 1 */
-	addib,COND(>)		-1, %arg2, fioneloop	/* Outer loop count decr */
-	fice,m		%arg1(%sr1, %arg0)	/* Fice for one loop */
+	/* Some implementations may flush with a single fice instruction */
+	cmpib,COND(>>=),n	15, %arg2, fioneloop2
+
+fioneloop1:
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	fice,m		%arg1(%sr4, %arg0)
+	addib,COND(>)	-16, %arg2, fioneloop1
+	fice,m		%arg1(%sr4, %arg0)
+
+	/* Check if done */
+	cmpb,COND(=),n	%arg2, %r0, fisync	/* Predict branch taken */
+
+fioneloop2:
+	addib,COND(>)	-1, %arg2, fioneloop2	/* Outer loop count decr */
+	fice,m		%arg1(%sr4, %arg0)	/* Fice for one loop */
 
 fisync:
 	sync
@@ -240,8 +265,7 @@ ENTRY(flush_data_cache_local)
 	.callinfo NO_CALLS
 	.entry
 
-	mtsp		%r0, %sr1
-	load32 		cache_info, %r1
+	load32		cache_info, %r1
 
 	/* Flush Data Cache */
 
@@ -249,20 +273,46 @@ ENTRY(flush_data_cache_local)
 	LDREG		DCACHE_STRIDE(%r1), %arg1
 	LDREG		DCACHE_COUNT(%r1), %arg2
 	LDREG		DCACHE_LOOP(%r1), %arg3
-	rsm		PSW_SM_I, %r22
+	rsm		PSW_SM_I, %r22		/* No mmgt ops during loop*/
 	addib,COND(=)		-1, %arg3, fdoneloop	/* Preadjust and test */
 	movb,<,n	%arg3, %r31, fdsync	/* If loop < 0, do sync */
 
 fdmanyloop:					/* Loop if LOOP >= 2 */
 	addib,COND(>)		-1, %r31, fdmanyloop	/* Adjusted inner loop decr */
-	fdce		%r0(%sr1, %arg0)
-	fdce,m		%arg1(%sr1, %arg0)	/* Last fdce and addr adjust */
+	fdce		%r0(%arg0)
+	fdce,m		%arg1(%arg0)	/* Last fdce and addr adjust */
 	movb,tr		%arg3, %r31, fdmanyloop	/* Re-init inner loop count */
 	addib,COND(<=),n	-1, %arg2, fdsync	/* Outer loop decr */
 
 fdoneloop:					/* Loop if LOOP = 1 */
-	addib,COND(>)		-1, %arg2, fdoneloop	/* Outer loop count decr */
-	fdce,m		%arg1(%sr1, %arg0)	/* Fdce for one loop */
+	/* Some implementations may flush with a single fdce instruction */
+	cmpib,COND(>>=),n	15, %arg2, fdoneloop2
+
+fdoneloop1:
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	fdce,m		%arg1(%arg0)
+	addib,COND(>)	-16, %arg2, fdoneloop1
+	fdce,m		%arg1(%arg0)
+
+	/* Check if done */
+	cmpb,COND(=),n	%arg2, %r0, fdsync	/* Predict branch taken */
+
+fdoneloop2:
+	addib,COND(>)	-1, %arg2, fdoneloop2	/* Outer loop count decr */
+	fdce,m		%arg1(%arg0)	/* Fdce for one loop */
 
 fdsync:
 	syncdma
@@ -277,7 +327,104 @@ ENDPROC(flush_data_cache_local)
 
 	.align	16
 
-ENTRY(copy_user_page_asm)
+/* Macros to serialize TLB purge operations on SMP.  */
+
+	.macro	tlb_lock	la,flags,tmp
+#ifdef CONFIG_SMP
+	ldil		L%pa_tlb_lock,%r1
+	ldo		R%pa_tlb_lock(%r1),\la
+	rsm		PSW_SM_I,\flags
+1:	LDCW		0(\la),\tmp
+	cmpib,<>,n	0,\tmp,3f
+2:	ldw		0(\la),\tmp
+	cmpb,<>		%r0,\tmp,1b
+	nop
+	b,n		2b
+3:
+#endif
+	.endm
+
+	.macro	tlb_unlock	la,flags,tmp
+#ifdef CONFIG_SMP
+	ldi		1,\tmp
+	stw		\tmp,0(\la)
+	mtsm		\flags
+#endif
+	.endm
+
+/* Clear page using kernel mapping.  */
+
+ENTRY(clear_page_asm)
+	.proc
+	.callinfo NO_CALLS
+	.entry
+
+#ifdef CONFIG_64BIT
+
+	/* Unroll the loop.  */
+	ldi		(PAGE_SIZE / 128), %r1
+
+1:
+	std		%r0, 0(%r26)
+	std		%r0, 8(%r26)
+	std		%r0, 16(%r26)
+	std		%r0, 24(%r26)
+	std		%r0, 32(%r26)
+	std		%r0, 40(%r26)
+	std		%r0, 48(%r26)
+	std		%r0, 56(%r26)
+	std		%r0, 64(%r26)
+	std		%r0, 72(%r26)
+	std		%r0, 80(%r26)
+	std		%r0, 88(%r26)
+	std		%r0, 96(%r26)
+	std		%r0, 104(%r26)
+	std		%r0, 112(%r26)
+	std		%r0, 120(%r26)
+
+	/* Note reverse branch hint for addib is taken.  */
+	addib,COND(>),n	-1, %r1, 1b
+	ldo		128(%r26), %r26
+
+#else
+
+	/*
+	 * Note that until (if) we start saving the full 64-bit register
+	 * values on interrupt, we can't use std on a 32 bit kernel.
+	 */
+	ldi		(PAGE_SIZE / 64), %r1
+
+1:
+	stw		%r0, 0(%r26)
+	stw		%r0, 4(%r26)
+	stw		%r0, 8(%r26)
+	stw		%r0, 12(%r26)
+	stw		%r0, 16(%r26)
+	stw		%r0, 20(%r26)
+	stw		%r0, 24(%r26)
+	stw		%r0, 28(%r26)
+	stw		%r0, 32(%r26)
+	stw		%r0, 36(%r26)
+	stw		%r0, 40(%r26)
+	stw		%r0, 44(%r26)
+	stw		%r0, 48(%r26)
+	stw		%r0, 52(%r26)
+	stw		%r0, 56(%r26)
+	stw		%r0, 60(%r26)
+
+	addib,COND(>),n	-1, %r1, 1b
+	ldo		64(%r26), %r26
+#endif
+	bv		%r0(%r2)
+	nop
+	.exit
+
+	.procend
+ENDPROC(clear_page_asm)
+
+/* Copy page using kernel mapping.  */
+
+ENTRY(copy_page_asm)
 	.proc
 	.callinfo NO_CALLS
 	.entry
@@ -285,18 +432,14 @@ ENTRY(copy_user_page_asm)
 #ifdef CONFIG_64BIT
 	/* PA8x00 CPUs can consume 2 loads or 1 store per cycle.
 	 * Unroll the loop by hand and arrange insn appropriately.
-	 * GCC probably can do this just as well.
+	 * Prefetch doesn't improve performance on rp3440.
+	 * GCC probably can do this just as well...
 	 */
 
-	ldd		0(%r25), %r19
 	ldi		(PAGE_SIZE / 128), %r1
 
-	ldw		64(%r25), %r0		/* prefetch 1 cacheline ahead */
-	ldw		128(%r25), %r0		/* prefetch 2 */
-
-1:	ldd		8(%r25), %r20
-	ldw		192(%r25), %r0		/* prefetch 3 */
-	ldw		256(%r25), %r0		/* prefetch 4 */
+1:	ldd		0(%r25), %r19
+	ldd		8(%r25), %r20
 
 	ldd		16(%r25), %r21
 	ldd		24(%r25), %r22
@@ -330,20 +473,16 @@ ENTRY(copy_user_page_asm)
 
 	ldd		112(%r25), %r21
 	ldd		120(%r25), %r22
+	ldo		128(%r25), %r25
 	std		%r19, 96(%r26)
 	std		%r20, 104(%r26)
 
-	ldo		128(%r25), %r25
 	std		%r21, 112(%r26)
 	std		%r22, 120(%r26)
-	ldo		128(%r26), %r26
 
-	/* conditional branches nullify on forward taken branch, and on
-	 * non-taken backward branch. Note that .+4 is a backwards branch.
-	 * The ldd should only get executed if the branch is taken.
-	 */
-	addib,COND(>),n	-1, %r1, 1b		/* bundle 10 */
-	ldd		0(%r25), %r19		/* start next loads */
+	/* Note reverse branch hint for addib is taken.  */
+	addib,COND(>),n	-1, %r1, 1b
+	ldo		128(%r26), %r26
 
 #else
 
@@ -399,7 +538,7 @@ ENTRY(copy_user_page_asm)
 	.exit
 
 	.procend
-ENDPROC(copy_user_page_asm)
+ENDPROC(copy_page_asm)
 
 /*
  * NOTE: Code in clear_user_page has a hard coded dependency on the
@@ -422,8 +561,6 @@ ENDPROC(copy_user_page_asm)
  *          %r23 physical page (shifted for tlb insert) of "from" translation
  */
 
-#if 0
-
 	/*
 	 * We can't do this since copy_user_page is used to bring in
 	 * file data that might have instructions. Since the data would
@@ -435,6 +572,7 @@ ENDPROC(copy_user_page_asm)
 	 * use it if more information is passed into copy_user_page().
 	 * Have to do some measurements to see if it is worthwhile to
 	 * lobby for such a change.
+	 *
 	 */
 
 ENTRY(copy_user_page_asm)
@@ -442,16 +580,21 @@ ENTRY(copy_user_page_asm)
 	.callinfo NO_CALLS
 	.entry
 
+	/* Convert virtual `to' and `from' addresses to physical addresses.
+	   Move `from' physical address to non shadowed register.  */
 	ldil		L%(__PAGE_OFFSET), %r1
 	sub		%r26, %r1, %r26
-	sub		%r25, %r1, %r23		/* move physical addr into non shadowed reg */
+	sub		%r25, %r1, %r23
 
 	ldil		L%(TMPALIAS_MAP_START), %r28
 	/* FIXME for different page sizes != 4k */
 #ifdef CONFIG_64BIT
-	extrd,u		%r26,56,32, %r26		/* convert phys addr to tlb insert format */
-	extrd,u		%r23,56,32, %r23		/* convert phys addr to tlb insert format */
-	depd		%r24,63,22, %r28		/* Form aliased virtual address 'to' */
+#if (TMPALIAS_MAP_START >= 0x80000000)
+	depdi		0, 31,32, %r28		/* clear any sign extension */
+#endif
+	extrd,u		%r26,56,32, %r26	/* convert phys addr to tlb insert format */
+	extrd,u		%r23,56,32, %r23	/* convert phys addr to tlb insert format */
+	depd		%r24,63,22, %r28	/* Form aliased virtual address 'to' */
 	depdi		0, 63,12, %r28		/* Clear any offset bits */
 	copy		%r28, %r29
 	depdi		1, 41,1, %r29		/* Form aliased virtual address 'from' */
@@ -466,10 +609,76 @@ ENTRY(copy_user_page_asm)
 
 	/* Purge any old translations */
 
+#ifdef CONFIG_PA20
+	pdtlb,l		0(%r28)
+	pdtlb,l		0(%r29)
+#else
+	tlb_lock	%r20,%r21,%r22
 	pdtlb		0(%r28)
 	pdtlb		0(%r29)
+	tlb_unlock	%r20,%r21,%r22
+#endif
 
-	ldi		64, %r1
+#ifdef CONFIG_64BIT
+	/* PA8x00 CPUs can consume 2 loads or 1 store per cycle.
+	 * Unroll the loop by hand and arrange insn appropriately.
+	 * GCC probably can do this just as well.
+	 */
+
+	ldd		0(%r29), %r19
+	ldi		(PAGE_SIZE / 128), %r1
+
+1:	ldd		8(%r29), %r20
+
+	ldd		16(%r29), %r21
+	ldd		24(%r29), %r22
+	std		%r19, 0(%r28)
+	std		%r20, 8(%r28)
+
+	ldd		32(%r29), %r19
+	ldd		40(%r29), %r20
+	std		%r21, 16(%r28)
+	std		%r22, 24(%r28)
+
+	ldd		48(%r29), %r21
+	ldd		56(%r29), %r22
+	std		%r19, 32(%r28)
+	std		%r20, 40(%r28)
+
+	ldd		64(%r29), %r19
+	ldd		72(%r29), %r20
+	std		%r21, 48(%r28)
+	std		%r22, 56(%r28)
+
+	ldd		80(%r29), %r21
+	ldd		88(%r29), %r22
+	std		%r19, 64(%r28)
+	std		%r20, 72(%r28)
+
+	ldd		 96(%r29), %r19
+	ldd		104(%r29), %r20
+	std		%r21, 80(%r28)
+	std		%r22, 88(%r28)
+
+	ldd		112(%r29), %r21
+	ldd		120(%r29), %r22
+	std		%r19, 96(%r28)
+	std		%r20, 104(%r28)
+
+	ldo		128(%r29), %r29
+	std		%r21, 112(%r28)
+	std		%r22, 120(%r28)
+	ldo		128(%r28), %r28
+
+	/* conditional branches nullify on forward taken branch, and on
+	 * non-taken backward branch. Note that .+4 is a backwards branch.
+	 * The ldd should only get executed if the branch is taken.
+	 */
+	addib,COND(>),n	-1, %r1, 1b		/* bundle 10 */
+	ldd		0(%r29), %r19		/* start next loads */
+
+#else
+	ldi		(PAGE_SIZE / 64), %r1
 
 	/*
 	 * This loop is optimized for PCXL/PCXL2 ldw/ldw and stw/stw
@@ -480,9 +689,7 @@ ENTRY(copy_user_page_asm)
 	 * use ldd/std on a 32 bit kernel.
 	 */
 
-
-1:
-	ldw		0(%r29), %r19
+1:	ldw		0(%r29), %r19
 	ldw		4(%r29), %r20
 	ldw		8(%r29), %r21
 	ldw		12(%r29), %r22
@@ -515,8 +722,10 @@ ENTRY(copy_user_page_asm)
 	stw		%r21, 56(%r28)
 	stw		%r22, 60(%r28)
 	ldo		64(%r28), %r28
+
 	addib,COND(>)		-1, %r1,1b
 	ldo		64(%r29), %r29
+#endif
 
 	bv		%r0(%r2)
 	nop
@@ -524,9 +733,8 @@ ENTRY(copy_user_page_asm)
 
 	.procend
 ENDPROC(copy_user_page_asm)
-#endif
 
-ENTRY(__clear_user_page_asm)
+ENTRY(clear_user_page_asm)
 	.proc
 	.callinfo NO_CALLS
 	.entry
@@ -550,7 +758,13 @@ ENTRY(__clear_user_page_asm)
 
 	/* Purge any old translation */
 
+#ifdef CONFIG_PA20
+	pdtlb,l		0(%r28)
+#else
+	tlb_lock	%r20,%r21,%r22
 	pdtlb		0(%r28)
+	tlb_unlock	%r20,%r21,%r22
+#endif
 
 #ifdef CONFIG_64BIT
 	ldi		(PAGE_SIZE / 128), %r1
@@ -580,8 +794,7 @@ ENTRY(__clear_user_page_asm)
 #else	/* ! CONFIG_64BIT */
 	ldi		(PAGE_SIZE / 64), %r1
 
-1:
-	stw		%r0, 0(%r28)
+1:	stw		%r0, 0(%r28)
 	stw		%r0, 4(%r28)
 	stw		%r0, 8(%r28)
 	stw		%r0, 12(%r28)
@@ -606,7 +819,7 @@ ENTRY(__clear_user_page_asm)
 	.exit
 
 	.procend
-ENDPROC(__clear_user_page_asm)
+ENDPROC(clear_user_page_asm)
 
 ENTRY(flush_dcache_page_asm)
 	.proc
@@ -630,7 +843,13 @@ ENTRY(flush_dcache_page_asm)
 
 	/* Purge any old translation */
 
+#ifdef CONFIG_PA20
+	pdtlb,l		0(%r28)
+#else
+	tlb_lock	%r20,%r21,%r22
 	pdtlb		0(%r28)
+	tlb_unlock	%r20,%r21,%r22
+#endif
 
 	ldil		L%dcache_stride, %r1
 	ldw		R%dcache_stride(%r1), %r1
@@ -663,8 +882,17 @@ ENTRY(flush_dcache_page_asm)
 	fdc,m		%r1(%r28)
 
 	sync
+
+#ifdef CONFIG_PA20
+	pdtlb,l		0(%r25)
+#else
+	tlb_lock	%r20,%r21,%r22
+	pdtlb		0(%r25)
+	tlb_unlock	%r20,%r21,%r22
+#endif
+
 	bv		%r0(%r2)
-	pdtlb		(%r25)
+	nop
 	.exit
 
 	.procend
@@ -692,7 +920,13 @@ ENTRY(flush_icache_page_asm)
 
 	/* Purge any old translation */
 
-	pitlb		(%sr4,%r28)
+#ifdef CONFIG_PA20
+	pitlb,l         %r0(%sr4,%r28)
+#else
+	tlb_lock        %r20,%r21,%r22
+	pitlb           (%sr4,%r28)
+	tlb_unlock      %r20,%r21,%r22
+#endif
 
 	ldil		L%icache_stride, %r1
 	ldw		R%icache_stride(%r1), %r1
@@ -727,8 +961,17 @@ ENTRY(flush_icache_page_asm)
 	fic,m		%r1(%sr4,%r28)
 
 	sync
+
+#ifdef CONFIG_PA20
+	pitlb,l         %r0(%sr4,%r25)
+#else
+	tlb_lock        %r20,%r21,%r22
+	pitlb           (%sr4,%r25)
+	tlb_unlock      %r20,%r21,%r22
+#endif
+
 	bv		%r0(%r2)
-	pitlb		(%sr4,%r25)
+	nop
 	.exit
 
 	.procend
@@ -777,7 +1020,7 @@ ENTRY(flush_kernel_dcache_page_asm)
 	.procend
 ENDPROC(flush_kernel_dcache_page_asm)
 
-ENTRY(purge_kernel_dcache_page)
+ENTRY(purge_kernel_dcache_page_asm)
 	.proc
 	.callinfo NO_CALLS
 	.entry
@@ -817,7 +1060,7 @@ ENTRY(purge_kernel_dcache_page)
 	.exit
 
 	.procend
-ENDPROC(purge_kernel_dcache_page)
+ENDPROC(purge_kernel_dcache_page_asm)
 
 ENTRY(flush_user_dcache_range_asm)
 	.proc
diff --git a/arch/parisc/kernel/parisc_ksyms.c b/arch/parisc/kernel/parisc_ksyms.c
index ceec85d..6795dc6 100644
--- a/arch/parisc/kernel/parisc_ksyms.c
+++ b/arch/parisc/kernel/parisc_ksyms.c
@@ -157,5 +157,6 @@ extern void _mcount(void);
 EXPORT_SYMBOL(_mcount);
 #endif
 
-/* from pacache.S -- needed for copy_page */
-EXPORT_SYMBOL(copy_user_page_asm);
+/* from pacache.S -- needed for clear/copy_page */
+EXPORT_SYMBOL(clear_page_asm);
+EXPORT_SYMBOL(copy_page_asm);
diff --git a/arch/parisc/kernel/ptrace.c b/arch/parisc/kernel/ptrace.c
index 857c2f5..534abd4 100644
--- a/arch/parisc/kernel/ptrace.c
+++ b/arch/parisc/kernel/ptrace.c
@@ -26,7 +26,7 @@
 #include <asm/asm-offsets.h>
 
 /* PSW bits we allow the debugger to modify */
-#define USER_PSW_BITS	(PSW_N | PSW_V | PSW_CB)
+#define USER_PSW_BITS	(PSW_N | PSW_B | PSW_V | PSW_CB)
 
 /*
  * Called by kernel/ptrace.c when detaching..
diff --git a/arch/parisc/kernel/signal.c b/arch/parisc/kernel/signal.c
index 5379969..52c85b2 100644
--- a/arch/parisc/kernel/signal.c
+++ b/arch/parisc/kernel/signal.c
@@ -190,8 +190,10 @@ get_sigframe(struct k_sigaction *ka, unsigned long sp, size_t frame_size)
 	DBG(1,"get_sigframe: ka = %#lx, sp = %#lx, frame_size = %#lx\n",
 			(unsigned long)ka, sp, frame_size);
 	
+	/* Align alternate stack and reserve 64 bytes for the signal
+	   handler's frame marker.  */
 	if ((ka->sa.sa_flags & SA_ONSTACK) != 0 && ! sas_ss_flags(sp))
-		sp = current->sas_ss_sp; /* Stacks grow up! */
+		sp = (current->sas_ss_sp + 0x7f) & ~0x3f; /* Stacks grow up! */
 
 	DBG(1,"get_sigframe: Returning sp = %#lx\n", (unsigned long)sp);
 	return (void __user *) sp; /* Stacks grow up.  Fun. */
@@ -310,7 +312,7 @@ setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
 #if DEBUG_SIG
 	/* Assert that we're flushing in the correct space... */
 	{
-		int sid;
+		unsigned long sid;
 		asm ("mfsp %%sr3,%0" : "=r" (sid));
 		DBG(1,"setup_rt_frame: Flushing 64 bytes at space %#x offset %p\n",
 		       sid, frame->tramp);
diff --git a/arch/parisc/kernel/sys_parisc.c b/arch/parisc/kernel/sys_parisc.c
index f76c108..a047b32 100644
--- a/arch/parisc/kernel/sys_parisc.c
+++ b/arch/parisc/kernel/sys_parisc.c
@@ -94,11 +94,12 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr,
 {
 	if (len > TASK_SIZE)
 		return -ENOMEM;
-	/* Might want to check for cache aliasing issues for MAP_FIXED case
-	 * like ARM or MIPS ??? --BenH.
-	 */
-	if (flags & MAP_FIXED)
+	if (flags & MAP_FIXED) {
+		if ((flags & MAP_SHARED) &&
+		    (addr - (pgoff << PAGE_SHIFT)) & (SHMLBA - 1))
+			return -EINVAL;
 		return addr;
+	}
 	if (!addr)
 		addr = TASK_UNMAPPED_BASE;
 
diff --git a/arch/parisc/kernel/syscall.S b/arch/parisc/kernel/syscall.S
index 86742df..c893ec4 100644
--- a/arch/parisc/kernel/syscall.S
+++ b/arch/parisc/kernel/syscall.S
@@ -309,10 +309,13 @@ tracesys_next:
 	LDREG   TASK_PT_GR25(%r1), %r25
 	LDREG   TASK_PT_GR24(%r1), %r24
 	LDREG   TASK_PT_GR23(%r1), %r23
-#ifdef CONFIG_64BIT
 	LDREG   TASK_PT_GR22(%r1), %r22
 	LDREG   TASK_PT_GR21(%r1), %r21
+#ifdef CONFIG_64BIT
 	ldo	-16(%r30),%r29			/* Reference param save area */
+#else
+	stw	%r22, -52(%r30)			/* 5th argument */
+	stw	%r21, -56(%r30)			/* 6th argument */
 #endif
 
 	comiclr,>>=	__NR_Linux_syscalls, %r20, %r0
diff --git a/arch/parisc/kernel/time.c b/arch/parisc/kernel/time.c
index 70e105d..4a24ba7 100644
--- a/arch/parisc/kernel/time.c
+++ b/arch/parisc/kernel/time.c
@@ -77,7 +77,7 @@ irqreturn_t __irq_entry timer_interrupt(int irq, void *dev_id)
 
 	cycles_elapsed = now - next_tick;
 
-	if ((cycles_elapsed >> 6) < cpt) {
+	if ((cycles_elapsed >> 7) < cpt) {
 		/* use "cheap" math (add/subtract) instead
 		 * of the more expensive div/mul method
 		 */
diff --git a/arch/parisc/math-emu/cnv_float.h b/arch/parisc/math-emu/cnv_float.h
index 9071e09..37299c7 100644
--- a/arch/parisc/math-emu/cnv_float.h
+++ b/arch/parisc/math-emu/cnv_float.h
@@ -347,16 +347,15 @@
     Sgl_isinexact_to_fix(sgl_value,exponent)
 
 #define Duint_from_sgl_mantissa(sgl_value,exponent,dresultA,dresultB)	\
-  {Sall(sgl_value) <<= SGL_EXP_LENGTH;  /*  left-justify  */		\
+  {unsigned int val = Sall(sgl_value) << SGL_EXP_LENGTH;		\
     if (exponent <= 31) {						\
     	Dintp1(dresultA) = 0;						\
-    	Dintp2(dresultB) = (unsigned)Sall(sgl_value) >> (31 - exponent); \
+    	Dintp2(dresultB) = val >> (31 - exponent);			\
     }									\
     else {								\
-    	Dintp1(dresultA) = Sall(sgl_value) >> (63 - exponent);		\
-    	Dintp2(dresultB) = Sall(sgl_value) << (exponent - 31);		\
+    	Dintp1(dresultA) = val >> (63 - exponent);			\
+    	Dintp2(dresultB) = exponent <= 62 ? val << (exponent - 31) : 0;	\
     }									\
-    Sall(sgl_value) >>= SGL_EXP_LENGTH;  /* return to original */	\
   }
 
 #define Duint_setzero(dresultA,dresultB) 	\
diff --git a/arch/parisc/mm/fault.c b/arch/parisc/mm/fault.c
index 18162ce..524bf5a 100644
--- a/arch/parisc/mm/fault.c
+++ b/arch/parisc/mm/fault.c
@@ -175,10 +175,12 @@ void do_page_fault(struct pt_regs *regs, unsigned long code,
 	struct mm_struct *mm = tsk->mm;
 	unsigned long acc_type;
 	int fault;
+	unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
 
 	if (in_atomic() || !mm)
 		goto no_context;
 
+retry:
 	down_read(&mm->mmap_sem);
 	vma = find_vma_prev(mm, address, &prev_vma);
 	if (!vma || address < vma->vm_start)
@@ -201,7 +203,12 @@ good_area:
 	 * fault.
 	 */
 
-	fault = handle_mm_fault(mm, vma, address, (acc_type & VM_WRITE) ? FAULT_FLAG_WRITE : 0);
+	fault = handle_mm_fault(mm, vma, address,
+		flags | ((acc_type & VM_WRITE) ? FAULT_FLAG_WRITE : 0));
+
+	if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
+		return;
+
 	if (unlikely(fault & VM_FAULT_ERROR)) {
 		/*
 		 * We hit a shared mapping outside of the file, or some
@@ -214,10 +221,22 @@ good_area:
 			goto bad_area;
 		BUG();
 	}
-	if (fault & VM_FAULT_MAJOR)
-		current->maj_flt++;
-	else
-		current->min_flt++;
+	if (flags & FAULT_FLAG_ALLOW_RETRY) {
+		if (fault & VM_FAULT_MAJOR)
+			current->maj_flt++;
+		else
+			current->min_flt++;
+		if (fault & VM_FAULT_RETRY) {
+			flags &= ~FAULT_FLAG_ALLOW_RETRY;
+
+			/* No need to up_read(&mm->mmap_sem) as we would
+			 * have already released it in __lock_page_or_retry
+			 * in mm/filemap.c.
+			 */
+
+			goto retry;
+		}
+	}
 	up_read(&mm->mmap_sem);
 	return;
 
diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c
index 57d86d0..f1b49d3 100644
--- a/kernel/irq/chip.c
+++ b/kernel/irq/chip.c
@@ -581,6 +581,7 @@ out_eoi:
 void
 handle_percpu_irq(unsigned int irq, struct irq_desc *desc)
 {
+	struct irqaction *action;
 	struct irq_chip *chip = irq_desc_get_chip(desc);
 
 	kstat_incr_irqs_this_cpu(irq, desc);
@@ -588,7 +589,9 @@ handle_percpu_irq(unsigned int irq, struct irq_desc *desc)
 	if (chip->irq_ack)
 		chip->irq_ack(&desc->irq_data);
 
-	handle_irq_event_percpu(desc, desc->action);
+	action = desc->action;
+	if (action)
+		handle_irq_event_percpu(desc, action);
 
 	if (chip->irq_eoi)
 		chip->irq_eoi(&desc->irq_data);
diff --git a/kernel/softirq.c b/kernel/softirq.c
index cc96bdc..6d68000 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -259,7 +259,7 @@ restart:
 		}
 		h++;
 		pending >>= 1;
-	} while (pending);
+	} while (pending && h >= (struct softirq_action *)0x1000);
 
 	local_irq_disable();
 

Re: Random crashes when calling git

[Rolf Eike Beer]

[-- Attachment #1: Type: text/plain, Size: 621 bytes --]

You wrote:
> On 31-Dec-12, at 6:03 AM, Rolf Eike Beer wrote:
> > I've put together the used git executable as well as 4 core dumps at
> > http://opensource.sf-tec.de/git-crash.tar.bz2
> > 
> > It would be nice if you could take a look and tell me if this gives
> > you
> > anything.
> 
> I looked at the first one:

> I think this is the infamous thread stack corruption bug.  There are
> numerous discussions about it on the list.
> 
> It's just a guess but I believe my kernel patch may help.  Attaching
> latest
> version for 3.7.1.

Thx, I'll have a look when I'm back in the office next week.

Until next year,

Eike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: Random crashes when calling git

[Rolf Eike Beer]

John David Anglin wrote:
> On 31-Dec-12, at 6:03 AM, Rolf Eike Beer wrote:
>
>> I've put together the used git executable as well as 4 core dumps at
>> http://opensource.sf-tec.de/git-crash.tar.bz2
>>
>> It would be nice if you could take a look and tell me if this gives  
>> you
>> anything.

> I think this is the infamous thread stack corruption bug.  There are
> numerous discussions about it on the list.
>
> It's just a guess but I believe my kernel patch may help.  Attaching  
> latest
> version for 3.7.1.
>
> If the patch doesn't help, then I think we need to review the glibc
> patch sets for debian and gentoo.  I have installed an unmodified
> version of 2.13-37.

I have now applied this patch and while at it added another 1GB of RAM, 
but I don't think the latter will make any difference.

I was now able to run the CMake git test for more than a dozen times in 
a row without a single failure and also the git testsuite is now in the 
range of the 6000 with it's tests, which is far beyond anything it was 
able to reach until now. So I think that your patch indeed makes things 
better.

Time to split it up and submit it upstream ;)

Eike

Re: Random crashes when calling git

[John David Anglin]

On 1/8/2013 9:30 AM, Rolf Eike Beer wrote:
> I was now able to run the CMake git test for more than a dozen times 
> in a row without a single failure and also the git testsuite is now in 
> the range of the 6000 with it's tests, which is far beyond anything it 
> was able to reach until now. So I think that your patch indeed makes 
> things better.
I have have built git several times and run the full testsuite successfully.
>
> Time to split it up and submit it upstream ;)
>
Helge has recently setup a new parisc-linux git tree on kernel.org and 
is going to help with
pushing patches upstream.  So, there is hope :-)

Dave

-- 
John David Anglin    dave.anglin@bell.net