gf-complete alignment constraints

gf-complete alignment constraints

[Loic Dachary]

[-- Attachment #1: Type: text/plain, Size: 1907 bytes --]

Hi Kevin,

I've hit the following stack trace (and I can reproduce it which is nice ;-)

#0  __GI_exit (status=1) at exit.c:104
#1  0x00007ffff18f3312 in gf_set_region_data (rd=0x7fffd9b8b0a0, gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, bytes=2048, val=143, xor=1, align=16) at erasure-code/jerasure/gf-complete/src/gf.c:817
#2  0x00007ffff1932364 in gf_w8_split_multiply_region_sse (gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, val=143, bytes=2048, xor=1) at erasure-code/jerasure/gf-complete/src/gf_w8.c:1071
#3  0x00007ffff18db9cc in galois_w08_region_multiply (region=0x4a7b31f "", multby=143, nbytes=2048, r2=0x4d25800 "root:x:0:\ndaemon:x:1:\nbin:x:2:\nsys:x:3:\nadm:x:4:loic,swift,syslog\ntty:x:5:\ndisk:x:6:\nlp:x:7:\nmail:x:8:\nnews:x:9:\nuucp:x:10:\nman:x:12:\nproxy:x:13:\nkmem:x:15:\ndialout:x:20:\nfax:x:21:\nvoice:x:22:\ncdrom:x"..., add=1) at erasure-code/jerasure/jerasure/src/galois.c:295
#4  0x00007ffff18ddb42 in jerasure_matrix_dotprod (k=2, w=8, matrix_row=0x47fa558, src_ids=0x0, dest_id=3, data_ptrs=0x7fffd9b8b260, coding_ptrs=0x7fffd9b8b270, size=2048) at erasure-code/jerasure/jerasure/src/jerasure.c:626

https://github.com/ceph/gf-complete/blob/v1/src/gf.c#L811 indeed checks that src=0x4a7b31f, dest=0x4d25800 are aligned and finds they are not. If I'm not mistaken src comes from https://github.com/ceph/jerasure/blob/v2/src/galois.c#L294 which will allocate it via https://github.com/ceph/jerasure/blob/v2/src/galois.c#L292 i.e https://github.com/ceph/jerasure/blob/v2/src/galois.c#L191 and then https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 which does a malloc that does not enforce alignment.

Should we relax the requirements or make sure https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 uses a logic similar to https://github.com/ceph/ceph/blob/giant/src/common/buffer.cc#L235 ?

Cheers
-- 
Loïc Dachary, Artisan Logiciel Libre


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: gf-complete alignment constraints

[Loic Dachary]

[-- Attachment #1: Type: text/plain, Size: 2383 bytes --]



On 01/12/2014 22:01, Kevin Greenan wrote:
> Hey Loic,
> 
> Makes sense.  I can have a look tonight.

I'm actually mistaken, confused the first argument with the second one :-/ I'll keep looking.

> 
> Thanks!
> -kevin
> 
> On Mon, Dec 1, 2014 at 11:15 AM, Loic Dachary <loic@dachary.org <mailto:loic@dachary.org>> wrote:
> 
>     Hi Kevin,
> 
>     I've hit the following stack trace (and I can reproduce it which is nice ;-)
> 
>     #0  __GI_exit (status=1) at exit.c:104
>     #1  0x00007ffff18f3312 in gf_set_region_data (rd=0x7fffd9b8b0a0, gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, bytes=2048, val=143, xor=1, align=16) at erasure-code/jerasure/gf-complete/src/gf.c:817
>     #2  0x00007ffff1932364 in gf_w8_split_multiply_region_sse (gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, val=143, bytes=2048, xor=1) at erasure-code/jerasure/gf-complete/src/gf_w8.c:1071
>     #3  0x00007ffff18db9cc in galois_w08_region_multiply (region=0x4a7b31f "", multby=143, nbytes=2048, r2=0x4d25800 "root:x:0:\ndaemon:x:1:\nbin:x:2:\nsys:x:3:\nadm:x:4:loic,swift,syslog\ntty:x:5:\ndisk:x:6:\nlp:x:7:\nmail:x:8:\nnews:x:9:\nuucp:x:10:\nman:x:12:\nproxy:x:13:\nkmem:x:15:\ndialout:x:20:\nfax:x:21:\nvoice:x:22:\ncdrom:x"..., add=1) at erasure-code/jerasure/jerasure/src/galois.c:295
>     #4  0x00007ffff18ddb42 in jerasure_matrix_dotprod (k=2, w=8, matrix_row=0x47fa558, src_ids=0x0, dest_id=3, data_ptrs=0x7fffd9b8b260, coding_ptrs=0x7fffd9b8b270, size=2048) at erasure-code/jerasure/jerasure/src/jerasure.c:626
> 
>     https://github.com/ceph/gf-complete/blob/v1/src/gf.c#L811 indeed checks that src=0x4a7b31f, dest=0x4d25800 are aligned and finds they are not. If I'm not mistaken src comes from https://github.com/ceph/jerasure/blob/v2/src/galois.c#L294 which will allocate it via https://github.com/ceph/jerasure/blob/v2/src/galois.c#L292 i.e https://github.com/ceph/jerasure/blob/v2/src/galois.c#L191 and then https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 which does a malloc that does not enforce alignment.
> 
>     Should we relax the requirements or make sure https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 uses a logic similar to https://github.com/ceph/ceph/blob/giant/src/common/buffer.cc#L235 ?
> 
>     Cheers
>     --
>     Loïc Dachary, Artisan Logiciel Libre
> 
> 

-- 
Loïc Dachary, Artisan Logiciel Libre


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: gf-complete alignment constraints

[Loic Dachary]

[-- Attachment #1: Type: text/plain, Size: 2562 bytes --]

You can ignore this thread, the problem is not in gf-complete or jerasure ;-)

Sorry for the noise

On 01/12/2014 22:47, Loic Dachary wrote:
> 
> 
> On 01/12/2014 22:01, Kevin Greenan wrote:
>> Hey Loic,
>>
>> Makes sense.  I can have a look tonight.
> 
> I'm actually mistaken, confused the first argument with the second one :-/ I'll keep looking.
> 
>>
>> Thanks!
>> -kevin
>>
>> On Mon, Dec 1, 2014 at 11:15 AM, Loic Dachary <loic@dachary.org <mailto:loic@dachary.org>> wrote:
>>
>>     Hi Kevin,
>>
>>     I've hit the following stack trace (and I can reproduce it which is nice ;-)
>>
>>     #0  __GI_exit (status=1) at exit.c:104
>>     #1  0x00007ffff18f3312 in gf_set_region_data (rd=0x7fffd9b8b0a0, gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, bytes=2048, val=143, xor=1, align=16) at erasure-code/jerasure/gf-complete/src/gf.c:817
>>     #2  0x00007ffff1932364 in gf_w8_split_multiply_region_sse (gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, val=143, bytes=2048, xor=1) at erasure-code/jerasure/gf-complete/src/gf_w8.c:1071
>>     #3  0x00007ffff18db9cc in galois_w08_region_multiply (region=0x4a7b31f "", multby=143, nbytes=2048, r2=0x4d25800 "root:x:0:\ndaemon:x:1:\nbin:x:2:\nsys:x:3:\nadm:x:4:loic,swift,syslog\ntty:x:5:\ndisk:x:6:\nlp:x:7:\nmail:x:8:\nnews:x:9:\nuucp:x:10:\nman:x:12:\nproxy:x:13:\nkmem:x:15:\ndialout:x:20:\nfax:x:21:\nvoice:x:22:\ncdrom:x"..., add=1) at erasure-code/jerasure/jerasure/src/galois.c:295
>>     #4  0x00007ffff18ddb42 in jerasure_matrix_dotprod (k=2, w=8, matrix_row=0x47fa558, src_ids=0x0, dest_id=3, data_ptrs=0x7fffd9b8b260, coding_ptrs=0x7fffd9b8b270, size=2048) at erasure-code/jerasure/jerasure/src/jerasure.c:626
>>
>>     https://github.com/ceph/gf-complete/blob/v1/src/gf.c#L811 indeed checks that src=0x4a7b31f, dest=0x4d25800 are aligned and finds they are not. If I'm not mistaken src comes from https://github.com/ceph/jerasure/blob/v2/src/galois.c#L294 which will allocate it via https://github.com/ceph/jerasure/blob/v2/src/galois.c#L292 i.e https://github.com/ceph/jerasure/blob/v2/src/galois.c#L191 and then https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 which does a malloc that does not enforce alignment.
>>
>>     Should we relax the requirements or make sure https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 uses a logic similar to https://github.com/ceph/ceph/blob/giant/src/common/buffer.cc#L235 ?
>>
>>     Cheers
>>     --
>>     Loïc Dachary, Artisan Logiciel Libre
>>
>>
> 

-- 
Loïc Dachary, Artisan Logiciel Libre


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: gf-complete alignment constraints

[Loic Dachary]

[-- Attachment #1: Type: text/plain, Size: 3529 bytes --]



On 01/12/2014 23:36, Kevin Greenan wrote:
> No problem.  Let me know if you need any help!

Thanks ! When investigating this problem I stumbled upon exit(1) in gf-complete and came up with a patch to replace them with assert(0):

https://github.com/ceph/gf-complete/pull/1/files

When exit(1) is called Ceph just stops with no useful information (the stderr does not in the ceph logs) while assert(0) is caught and shows a nice stack trace. It's not high priority though ;-)

Cheers

> 
> -kevin
> 
> On Mon, Dec 1, 2014 at 2:31 PM, Loic Dachary <loic@dachary.org <mailto:loic@dachary.org>> wrote:
> 
>     You can ignore this thread, the problem is not in gf-complete or jerasure ;-)
> 
>     Sorry for the noise
> 
>     On 01/12/2014 22:47, Loic Dachary wrote:
>     >
>     >
>     > On 01/12/2014 22:01, Kevin Greenan wrote:
>     >> Hey Loic,
>     >>
>     >> Makes sense.  I can have a look tonight.
>     >
>     > I'm actually mistaken, confused the first argument with the second one :-/ I'll keep looking.
>     >
>     >>
>     >> Thanks!
>     >> -kevin
>     >>
>     >> On Mon, Dec 1, 2014 at 11:15 AM, Loic Dachary <loic@dachary.org <mailto:loic@dachary.org> <mailto:loic@dachary.org <mailto:loic@dachary.org>>> wrote:
>     >>
>     >>     Hi Kevin,
>     >>
>     >>     I've hit the following stack trace (and I can reproduce it which is nice ;-)
>     >>
>     >>     #0  __GI_exit (status=1) at exit.c:104
>     >>     #1  0x00007ffff18f3312 in gf_set_region_data (rd=0x7fffd9b8b0a0, gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, bytes=2048, val=143, xor=1, align=16) at erasure-code/jerasure/gf-complete/src/gf.c:817
>     >>     #2  0x00007ffff1932364 in gf_w8_split_multiply_region_sse (gf=0x485dd40, src=0x4a7b31f, dest=0x4d25800, val=143, bytes=2048, xor=1) at erasure-code/jerasure/gf-complete/src/gf_w8.c:1071
>     >>     #3  0x00007ffff18db9cc in galois_w08_region_multiply (region=0x4a7b31f "", multby=143, nbytes=2048, r2=0x4d25800 "root:x:0:\ndaemon:x:1:\nbin:x:2:\nsys:x:3:\nadm:x:4:loic,swift,syslog\ntty:x:5:\ndisk:x:6:\nlp:x:7:\nmail:x:8:\nnews:x:9:\nuucp:x:10:\nman:x:12:\nproxy:x:13:\nkmem:x:15:\ndialout:x:20:\nfax:x:21:\nvoice:x:22:\ncdrom:x"..., add=1) at erasure-code/jerasure/jerasure/src/galois.c:295
>     >>     #4  0x00007ffff18ddb42 in jerasure_matrix_dotprod (k=2, w=8, matrix_row=0x47fa558, src_ids=0x0, dest_id=3, data_ptrs=0x7fffd9b8b260, coding_ptrs=0x7fffd9b8b270, size=2048) at erasure-code/jerasure/jerasure/src/jerasure.c:626
>     >>
>     >>     https://github.com/ceph/gf-complete/blob/v1/src/gf.c#L811 indeed checks that src=0x4a7b31f, dest=0x4d25800 are aligned and finds they are not. If I'm not mistaken src comes from https://github.com/ceph/jerasure/blob/v2/src/galois.c#L294 which will allocate it via https://github.com/ceph/jerasure/blob/v2/src/galois.c#L292 i.e https://github.com/ceph/jerasure/blob/v2/src/galois.c#L191 and then https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 which does a malloc that does not enforce alignment.
>     >>
>     >>     Should we relax the requirements or make sure https://github.com/ceph/jerasure/blob/v2/src/galois.c#L175 uses a logic similar to https://github.com/ceph/ceph/blob/giant/src/common/buffer.cc#L235 ?
>     >>
>     >>     Cheers
>     >>     --
>     >>     Loïc Dachary, Artisan Logiciel Libre
>     >>
>     >>
>     >
> 
>     --
>     Loïc Dachary, Artisan Logiciel Libre
> 
> 

-- 
Loïc Dachary, Artisan Logiciel Libre


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]