* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
@ 2007-10-11 21:42 ` David Miller
2007-10-11 22:37 ` David Miller
` (7 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: David Miller @ 2007-10-11 21:42 UTC (permalink / raw)
To: sparclinux
From: BERTRAND_Joël <joel.bertrand@systella.fr>
Date: Thu, 11 Oct 2007 20:44:56 +0200
> I have built a 2.6.23 kernel on a T1000 server. I have seen this
> message on system console :
What compiler did you use to build this and what debian based
distribution are you running?
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
2007-10-11 21:42 ` David Miller
@ 2007-10-11 22:37 ` David Miller
2007-10-12 5:47 ` BERTRAND Joël
` (6 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: David Miller @ 2007-10-11 22:37 UTC (permalink / raw)
To: sparclinux
From: BERTRAND_Joël <joel.bertrand@systella.fr>
Date: Thu, 11 Oct 2007 20:44:56 +0200
> Hello,
>
> I have built a 2.6.23 kernel on a T1000 server. I have seen this
> message on system console :
Looking at the trace some more, I'm %99.99999 sure you're using
gcc-4.2.x to build this and that compiler is known to miscompile SMP
sparc64 kernels.
gcc-4.1.x would never inline __flush_tsb_one() into flush_tsb_user(),
yet as is evident in your backtraces this is exactly what has
happened, therefore you must be using gcc-4.2.x or another non-4.1.x
compiler to build this.
Please rebuild with gcc-4.1.x, I guarentee the crash will go away.
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
2007-10-11 21:42 ` David Miller
2007-10-11 22:37 ` David Miller
@ 2007-10-12 5:47 ` BERTRAND Joël
2007-10-12 7:04 ` BERTRAND Joël
` (5 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: BERTRAND Joël @ 2007-10-12 5:47 UTC (permalink / raw)
To: sparclinux
David Miller wrote:
> From: BERTRAND_Joël <joel.bertrand@systella.fr>
> Date: Thu, 11 Oct 2007 20:44:56 +0200
>
>> Hello,
>>
>> I have built a 2.6.23 kernel on a T1000 server. I have seen this
>> message on system console :
David,
> Looking at the trace some more, I'm %99.99999 sure you're using
> gcc-4.2.x to build this and that compiler is known to miscompile SMP
> sparc64 kernels.
I a first time, I though that is was this trouble, but I can see the
same bug with debian kernel provided by debian/testing (a 2.6.22
kernel). I don't know what compiler was used to build debian kernel.
> gcc-4.1.x would never inline __flush_tsb_one() into flush_tsb_user(),
> yet as is evident in your backtraces this is exactly what has
> happened, therefore you must be using gcc-4.2.x or another non-4.1.x
> compiler to build this.
OK, I will try to rebuild a kernel with gcc-4.1. Thanks for your help.
Regards,
JKB
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
` (2 preceding siblings ...)
2007-10-12 5:47 ` BERTRAND Joël
@ 2007-10-12 7:04 ` BERTRAND Joël
2007-10-12 8:37 ` David Miller
` (4 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: BERTRAND Joël @ 2007-10-12 7:04 UTC (permalink / raw)
To: sparclinux
BERTRAND Joël wrote:
> David Miller wrote:
>> From: BERTRAND_Joël <joel.bertrand@systella.fr>
>> Date: Thu, 11 Oct 2007 20:44:56 +0200
>>
>>> Hello,
>>>
>>> I have built a 2.6.23 kernel on a T1000 server. I have seen this
>>> message on system console :
>
> David,
>
>> Looking at the trace some more, I'm %99.99999 sure you're using
>> gcc-4.2.x to build this and that compiler is known to miscompile SMP
>> sparc64 kernels.
>
> I a first time, I though that is was this trouble, but I can see the
> same bug with debian kernel provided by debian/testing (a 2.6.22
> kernel). I don't know what compiler was used to build debian kernel.
>
>> gcc-4.1.x would never inline __flush_tsb_one() into flush_tsb_user(),
>> yet as is evident in your backtraces this is exactly what has
>> happened, therefore you must be using gcc-4.2.x or another non-4.1.x
>> compiler to build this.
>
> OK, I will try to rebuild a kernel with gcc-4.1. Thanks for your help.
David,
I have rebuild a 2.6.23 kernel with gcc-4.1. When I try to format
(ext3fs) a raid5 volume, I _allways_ obtain (I have tested four times):
Root gershwin:[~] > mkfs.ext3 /dev/md8
mke2fs 1.40.2 (12-Jul-2007)
Filesystem labelOS type: Linux
Block size@96 (log=2)
Fragment size@96 (log=2)
183091200 inodes, 366181424 blocks
18309071 blocks (5.00%) reserved for the super user
First data block=0
Maximum filesystem blocks=0
11175 block groups
32768 blocks per group, 32768 fragments per group
16384 inodes per group
Superblock backups stored on blocks:
32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632,
2654208,
4096000, 7962624, 11239424, 20480000, 23887872, 71663616,
78675968,
102400000, 214990848
Writing inode tables: Kernel unaligned access at TPC[56004c]
xor_niagara_4+0x5c/0x128
sun4v_data_access_exception: ADDR[000000000053d354] CTX[0000]
TYPE[000a], going.
\|/ ____ \|/
"@'/ .. \`@"
/_| \__/ |_\
\__U_/
md7_raid5(2676): Dax [#1]
TSTATE: 00000000e2001602 TPC: 000000000042bc60 TNPC: 000000000042bc64 Y:
00000000 Not tainted
TPC: <do_int_load+0x78/0xf0>
g0: fffff800fb84c000 g1: fffff800fb84f570 g2: 0000000000000400 g3:
000000000000001c
g4: fffff800fc4d3600 g5: fffff800020a8000 g6: fffff800fb84c000 g7:
fffff800fb84f4d0
o0: fffff800fb84f5d0 o1: 0000000000000010 o2: 000000000053d354 o3:
0000000000000000
o4: 00000000000000e2 o5: 0000000000000080 sp: fffff800fb84ea01 ret_pc:
0000000000435e6c
RPC: <kernel_unaligned_trap+0x194/0x520>
l0: 00000000000000e2 l1: fffff800fb84f5d0 l2: 000000000000001c l3:
0000000000000000
l4: 0000000000000010 l5: 00000000000000e2 l6: 000000000053d354 l7:
0000000080009000
i0: fffff800fb84f4d0 i1: 00000000f89fe010 i2: fffff800fc4d3600 i3:
0000000000000000
i4: 0000000000000034 i5: 000000000000000b i6: fffff800fb84ead1 i7:
00000000004290c0
I7: <sun4v_do_mna+0x88/0xa0>
Caller[00000000004290c0]: sun4v_do_mna+0x88/0xa0
Caller[0000000000406b78]: sun4v_mna+0x64/0x68
Caller[000000000053e674]: async_xor+0x4bc/0x5a0
Caller[000000000053d344]: xor_blocks+0x8c/0xe0
Caller[000000000053e674]: async_xor+0x4bc/0x5a0
Caller[00000000005ef328]: ops_run_prexor+0xd0/0xe0
Caller[00000000005efce4]: raid5_run_ops+0x52c/0x5c0
Caller[00000000005f01b8]: handle_stripe5+0x440/0x1340
Caller[00000000005f211c]: handle_stripe+0x24/0x13e0
Caller[00000000005f37c4]: raid5d+0x2ec/0x3c0
Caller[00000000005ff8f0]: md_thread+0x38/0x140
Caller[0000000000478b40]: kthread+0x48/0x80
Caller[00000000004273d0]: kernel_thread+0x38/0x60
Caller[0000000000478de0]: kthreadd+0x148/0x1c0
Instruction DUMP: 8538a000 1068001f c4720000 <c48aa000> c68aa001
8528b038 ce8aa002 8728f030 c28aa003
313/11175
To build this new kernel, I have modified main Makefile to use gcc-4.1
(variable CC) dans dmesg returns:
PROMLIB: Sun IEEE Boot Prom 'OBP 4.23.4 2006/08/04 20:45'
PROMLIB: Root node compatible: sun4v
Linux version 2.6.23 (root@gershwin) (gcc version 4.1.3 20070831
(prerelease) (Debian 4.1.2-16)) #2 SMP Fri Oct 12 08:34:39 CEST 2007
ARCH: SUN4V
Ethernet address: 00:14:4f:6f:59:fe
OF stdout device is: /virtual-devices@100/console@1
PROM: Built device tree with 74930 bytes of memory.
For information:
Root gershwin:[~] > cat /proc/mdstat
Personalities : [raid1] [raid6] [raid5] [raid4]
md8 : active raid1 md7[0]
1464725696 blocks [2/1] [U_]
md7 : active raid5 sdc1[0] sdg1[5] sdh1[4] sdf1[3] sde1[2] sdd1[1]
1464725760 blocks level 5, 64k chunk, algorithm 2 [6/6] [UUUUUU]
[>....................] resync = 1.1% (3471096/292945152)
finish!1.0mi
n speed"856K/sec
md6 : active raid1 sda1[0] sdb1[1]
7815552 blocks [2/2] [UU]
md5 : active raid1 sda8[0] sdb8[1]
14538752 blocks [2/2] [UU]
md4 : active raid1 sda7[0] sdb7[1]
4883648 blocks [2/2] [UU]
md3 : active raid1 sda6[0] sdb6[1]
9767424 blocks [2/2] [UU]
md2 : active raid1 sda5[0] sdb5[1]
29294400 blocks [2/2] [UU]
md1 : active raid1 sda2[0] sdb2[1]
489856 blocks [2/2] [UU]
md0 : active raid1 sdb4[1] sda4[0]
4883648 blocks [2/2] [UU]
unused devices: <none>
Root gershwin:[~] >
and /dev/md8 is a raid1 volume shared on network (by iscsi [seems to not
work] or nbd).
Regards,
JKB
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
` (3 preceding siblings ...)
2007-10-12 7:04 ` BERTRAND Joël
@ 2007-10-12 8:37 ` David Miller
2007-10-12 8:50 ` David Miller
` (3 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: David Miller @ 2007-10-12 8:37 UTC (permalink / raw)
To: sparclinux
From: BERTRAND_Joël <joel.bertrand@systella.fr>
Date: Fri, 12 Oct 2007 09:04:16 +0200
> I have rebuild a 2.6.23 kernel with gcc-4.1. When I try to format
> (ext3fs) a raid5 volume, I _allways_ obtain (I have tested four times):
That's a different bug which I will try to look into.
At least confirm that the other problem no longer occurs
as that's all I'm interested in at this point.
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
` (4 preceding siblings ...)
2007-10-12 8:37 ` David Miller
@ 2007-10-12 8:50 ` David Miller
2007-10-12 8:52 ` BERTRAND Joël
` (2 subsequent siblings)
8 siblings, 0 replies; 10+ messages in thread
From: David Miller @ 2007-10-12 8:50 UTC (permalink / raw)
To: sparclinux
From: BERTRAND_Joël <joel.bertrand@systella.fr>
Date: Fri, 12 Oct 2007 09:04:16 +0200
> Writing inode tables: Kernel unaligned access at TPC[56004c]
> xor_niagara_4+0x5c/0x128
> sun4v_data_access_exception: ADDR[000000000053d354] CTX[0000]
> TYPE[000a], going.
Please try this patch:
diff --git a/arch/sparc64/lib/xor.S b/arch/sparc64/lib/xor.S
index a79c888..f44f58f 100644
--- a/arch/sparc64/lib/xor.S
+++ b/arch/sparc64/lib/xor.S
@@ -491,12 +491,12 @@ xor_niagara_4: /* %o0=bytes, %o1Þst, %o2=src1, %o3=src2, %o4=src3 */
ldda [%i1 + 0x10] %asi, %i2 /* %i2/%i3 = src1 + 0x10 */
xor %g2, %i4, %g2
xor %g3, %i5, %g3
- ldda [%i7 + 0x10] %asi, %i4 /* %i4/%i5 = src2 + 0x10 */
+ ldda [%l7 + 0x10] %asi, %i4 /* %i4/%i5 = src2 + 0x10 */
xor %l0, %g2, %l0
xor %l1, %g3, %l1
stxa %l0, [%i0 + 0x00] %asi
stxa %l1, [%i0 + 0x08] %asi
- ldda [%i6 + 0x10] %asi, %g2 /* %g2/%g3 = src3 + 0x10 */
+ ldda [%l6 + 0x10] %asi, %g2 /* %g2/%g3 = src3 + 0x10 */
ldda [%i0 + 0x10] %asi, %l0 /* %l0/%l1 = dest + 0x10 */
xor %i4, %i2, %i4
@@ -504,12 +504,12 @@ xor_niagara_4: /* %o0=bytes, %o1Þst, %o2=src1, %o3=src2, %o4=src3 */
ldda [%i1 + 0x20] %asi, %i2 /* %i2/%i3 = src1 + 0x20 */
xor %g2, %i4, %g2
xor %g3, %i5, %g3
- ldda [%i7 + 0x20] %asi, %i4 /* %i4/%i5 = src2 + 0x20 */
+ ldda [%l7 + 0x20] %asi, %i4 /* %i4/%i5 = src2 + 0x20 */
xor %l0, %g2, %l0
xor %l1, %g3, %l1
stxa %l0, [%i0 + 0x10] %asi
stxa %l1, [%i0 + 0x18] %asi
- ldda [%i6 + 0x20] %asi, %g2 /* %g2/%g3 = src3 + 0x20 */
+ ldda [%l6 + 0x20] %asi, %g2 /* %g2/%g3 = src3 + 0x20 */
ldda [%i0 + 0x20] %asi, %l0 /* %l0/%l1 = dest + 0x20 */
xor %i4, %i2, %i4
@@ -517,12 +517,12 @@ xor_niagara_4: /* %o0=bytes, %o1Þst, %o2=src1, %o3=src2, %o4=src3 */
ldda [%i1 + 0x30] %asi, %i2 /* %i2/%i3 = src1 + 0x30 */
xor %g2, %i4, %g2
xor %g3, %i5, %g3
- ldda [%i7 + 0x30] %asi, %i4 /* %i4/%i5 = src2 + 0x30 */
+ ldda [%l7 + 0x30] %asi, %i4 /* %i4/%i5 = src2 + 0x30 */
xor %l0, %g2, %l0
xor %l1, %g3, %l1
stxa %l0, [%i0 + 0x20] %asi
stxa %l1, [%i0 + 0x28] %asi
- ldda [%i6 + 0x30] %asi, %g2 /* %g2/%g3 = src3 + 0x30 */
+ ldda [%l6 + 0x30] %asi, %g2 /* %g2/%g3 = src3 + 0x30 */
ldda [%i0 + 0x30] %asi, %l0 /* %l0/%l1 = dest + 0x30 */
prefetch [%i1 + 0x40], #one_read
^ permalink raw reply related [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
` (5 preceding siblings ...)
2007-10-12 8:50 ` David Miller
@ 2007-10-12 8:52 ` BERTRAND Joël
2007-10-12 9:16 ` BERTRAND Joël
2007-10-12 9:27 ` David Miller
8 siblings, 0 replies; 10+ messages in thread
From: BERTRAND Joël @ 2007-10-12 8:52 UTC (permalink / raw)
To: sparclinux
David Miller wrote:
> From: BERTRAND_Joël <joel.bertrand@systella.fr>
> Date: Fri, 12 Oct 2007 09:04:16 +0200
>
>> I have rebuild a 2.6.23 kernel with gcc-4.1. When I try to format
>> (ext3fs) a raid5 volume, I _allways_ obtain (I have tested four times):
>
> That's a different bug which I will try to look into.
This bug is very strange. I have done a mkfs on a raid6 (3 GB) wihtout
any trouble, but on a raid5 (1,5 GB), it allways crash at the same
inode. Same constatation on my second T1000, thus, it is not an hardware
problem ;-)
> At least confirm that the other problem no longer occurs
> as that's all I'm interested in at this point.
I don't see the last one.
Regards,
JKB
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
` (6 preceding siblings ...)
2007-10-12 8:52 ` BERTRAND Joël
@ 2007-10-12 9:16 ` BERTRAND Joël
2007-10-12 9:27 ` David Miller
8 siblings, 0 replies; 10+ messages in thread
From: BERTRAND Joël @ 2007-10-12 9:16 UTC (permalink / raw)
To: sparclinux
David Miller wrote:
> From: BERTRAND_Joël <joel.bertrand@systella.fr>
> Date: Fri, 12 Oct 2007 09:04:16 +0200
>
>> Writing inode tables: Kernel unaligned access at TPC[56004c]
>> xor_niagara_4+0x5c/0x128
>> sun4v_data_access_exception: ADDR[000000000053d354] CTX[0000]
>> TYPE[000a], going.
>
> Please try this patch:
>
> diff --git a/arch/sparc64/lib/xor.S b/arch/sparc64/lib/xor.S
> index a79c888..f44f58f 100644
> --- a/arch/sparc64/lib/xor.S
> +++ b/arch/sparc64/lib/xor.S
> @@ -491,12 +491,12 @@ xor_niagara_4: /* %o0=bytes, %o1Þst, %o2=src1, %o3=src2, %o4=src3 */
> ldda [%i1 + 0x10] %asi, %i2 /* %i2/%i3 = src1 + 0x10 */
> xor %g2, %i4, %g2
> xor %g3, %i5, %g3
> - ldda [%i7 + 0x10] %asi, %i4 /* %i4/%i5 = src2 + 0x10 */
> + ldda [%l7 + 0x10] %asi, %i4 /* %i4/%i5 = src2 + 0x10 */
> xor %l0, %g2, %l0
> xor %l1, %g3, %l1
> stxa %l0, [%i0 + 0x00] %asi
> stxa %l1, [%i0 + 0x08] %asi
> - ldda [%i6 + 0x10] %asi, %g2 /* %g2/%g3 = src3 + 0x10 */
> + ldda [%l6 + 0x10] %asi, %g2 /* %g2/%g3 = src3 + 0x10 */
> ldda [%i0 + 0x10] %asi, %l0 /* %l0/%l1 = dest + 0x10 */
>
> xor %i4, %i2, %i4
> @@ -504,12 +504,12 @@ xor_niagara_4: /* %o0=bytes, %o1Þst, %o2=src1, %o3=src2, %o4=src3 */
> ldda [%i1 + 0x20] %asi, %i2 /* %i2/%i3 = src1 + 0x20 */
> xor %g2, %i4, %g2
> xor %g3, %i5, %g3
> - ldda [%i7 + 0x20] %asi, %i4 /* %i4/%i5 = src2 + 0x20 */
> + ldda [%l7 + 0x20] %asi, %i4 /* %i4/%i5 = src2 + 0x20 */
> xor %l0, %g2, %l0
> xor %l1, %g3, %l1
> stxa %l0, [%i0 + 0x10] %asi
> stxa %l1, [%i0 + 0x18] %asi
> - ldda [%i6 + 0x20] %asi, %g2 /* %g2/%g3 = src3 + 0x20 */
> + ldda [%l6 + 0x20] %asi, %g2 /* %g2/%g3 = src3 + 0x20 */
> ldda [%i0 + 0x20] %asi, %l0 /* %l0/%l1 = dest + 0x20 */
>
> xor %i4, %i2, %i4
> @@ -517,12 +517,12 @@ xor_niagara_4: /* %o0=bytes, %o1Þst, %o2=src1, %o3=src2, %o4=src3 */
> ldda [%i1 + 0x30] %asi, %i2 /* %i2/%i3 = src1 + 0x30 */
> xor %g2, %i4, %g2
> xor %g3, %i5, %g3
> - ldda [%i7 + 0x30] %asi, %i4 /* %i4/%i5 = src2 + 0x30 */
> + ldda [%l7 + 0x30] %asi, %i4 /* %i4/%i5 = src2 + 0x30 */
> xor %l0, %g2, %l0
> xor %l1, %g3, %l1
> stxa %l0, [%i0 + 0x20] %asi
> stxa %l1, [%i0 + 0x28] %asi
> - ldda [%i6 + 0x30] %asi, %g2 /* %g2/%g3 = src3 + 0x30 */
> + ldda [%l6 + 0x30] %asi, %g2 /* %g2/%g3 = src3 + 0x30 */
> ldda [%i0 + 0x30] %asi, %l0 /* %l0/%l1 = dest + 0x30 */
>
> prefetch [%i1 + 0x40], #one_read
David,
I think that your patch fix this bug. I shall test on other T1000 this
afternoon.
Thanks,
JKB
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: sun4v_data_access_exception on new 2.6.23
2007-10-11 18:44 sun4v_data_access_exception on new 2.6.23 BERTRAND Joël
` (7 preceding siblings ...)
2007-10-12 9:16 ` BERTRAND Joël
@ 2007-10-12 9:27 ` David Miller
8 siblings, 0 replies; 10+ messages in thread
From: David Miller @ 2007-10-12 9:27 UTC (permalink / raw)
To: sparclinux
From: BERTRAND_Joël <joel.bertrand@systella.fr>
Date: Fri, 12 Oct 2007 11:16:11 +0200
> I think that your patch fix this bug. I shall test on other T1000 this
> afternoon.
Thank you for testing.
^ permalink raw reply [flat|nested] 10+ messages in thread