** le kernel 5.8 est dispo ** (page 70)

Il faut faire attention avec ces versions : ce sont vraiment des versions de tests. Lisez bien les changelog :wink:

The latest snapshot for the stable Linux kernel tree is: 2.6.7-rc1-bk3 2004-05-26 11:39 UTC

tient pour une foie c moi qui annonce le nouveau kernelvous avez vu sinon cette annonce :
http://www.toolinux.com/news/communaute/developpement_linux_linus_reclame_ordre_et_transparence_ar4898.html

de linus a propos du noyau

Ton lien affiche ceci :

[fixed]Syntaxe boucle incorrecte

BOUCLE_forums: double définition [/fixed]

bah na je viens de le tester il marche chez moi sinon c la premiere news de toolinux : http://www.toolinux.com/

Je ne fais pas l’annonce des bk car sinon on va avoir 15*10^6 questions à leur sujet ici et pleins de gens qi se plaignent que “ça marche mal”. C’est clair qu’il faut bien que certains teste ces noyaux mais il faut savoir à quoi s’attenndre :wink:

Ce message n’était pas conforme aux règles d’utilisation du nouveau forum :

:smiley:

ça peux pas etre pire que les love (rc + bk + mm + autres conneries très dangereuses a mon gout)

:love:

http://marc.theaimsgroup.com/?t=108564823300011&r=1&w=2

  • Various net driver updates

  • Significant rework of the RCU code core to fix serious scalability
    problems on huge SMP.

  • Devicemapper update

  • Various random other things

yaisse ça veut dire que le love est pas loin [:aol]

Ben de mon côté j’ai mis en évidence un deadlock entre client et serveurs NFSv4 si on utilise de nombreux client (avec nfs_fsstress du ltp)… Je m’en vais voir ce que je peux faire pour débogguer ça.

Bon courrage. J’espère que c’est systématique et que le bug ,e sera pas trop dur à isoler.

il a l’air très bien ce 2.6.7-rc1-mm1 :super:

12:17:51 up 1 day, 15:55,  7 users,  load average: 1.71, 1.57, 1.13

sans que le speedtouch USB ne plante :smiley:

le 2.6.7-rc2 :
http://www.kernel.org/pub/linux/kernel/v2.6/testing/ChangeLog-2.6.7-rc2

le 2.6.7-rc1 cooker plante des le demarrage chez moi avec une erreur incomprehensible avec plein d’adresses

http://marc.theaimsgroup.com/?l=linux-kernel&m=108590022702388&w=2

Principalement un update d’ALSA
“.0.5a driver release | bugfix for the “hang-up” problem in the ALSA driver initialization” dixit le site d’alsa : Je ne sais pas ce que c’est :wink:

Je vais mettre à jour les kernels non-officiels Fedora :slight_smile:

http://marc.theaimsgroup.com/?t=108608174900002&r=1&w=2

Ce 2.6.7rc2-mm1 déclanche bcp de réactions.

Ben je suis toujours sur mon bug NFSv4…

J’extrait de la liste de diffusion:
v_atekor:
As suite to the apparently deadlock I had Friday, I had updated the 2 kernels
into Linux 2.6.7-rc1 All citi patches on both client and server, and I try to
run nfs_fsstress again.

Quickly the shared directory looks to be freezed, and is not accessible by the
shell)

Here are the traces I had in the kernel logs.

I remind that nfs_fsstress:

  • mount 2 time the same remote directory on 2 mount point
  • launch 100 forked fssress process accessing on each mount point directory

Here are the traces for the client :

Jun 1 17:07:45 nfs2 kernel: decode_getfattr: xdr error 10008!
Jun 1 17:07:55 nfs2 sshd(pam_unix)[2233]: session opened for user root by
(uid=0)
Jun 1 17:08:00 nfs2 kernel: RESERVE_SPACE(3139) failed in function
encode_create
Jun 1 17:08:00 nfs2 kernel: ------------[ cut here ]------------
Jun 1 17:08:00 nfs2 kernel: kernel BUG at fs/nfs/nfs4xdr.c:626!
Jun 1 17:08:00 nfs2 kernel: invalid operand: 0000 [#1]
Jun 1 17:08:00 nfs2 kernel: PREEMPT SMP
Jun 1 17:08:00 nfs2 kernel: Modules linked in: autofs e1000
Jun 1 17:08:00 nfs2 kernel: CPU: 0
Jun 1 17:08:00 nfs2 kernel: EIP: 0060:[] Not tainted
Jun 1 17:08:00 nfs2 kernel: EFLAGS: 00010282 (2.6.7-rc1-CITI_NFS4_ALL-1)
Jun 1 17:08:01 nfs2 kernel: EIP is at encode_create+0x142/0x1c5
Jun 1 17:08:01 nfs2 kernel: eax: 00000035 ebx: 00000000 ecx: 00000002
edx: c03f023c
Jun 1 17:08:01 nfs2 kernel: esi: f09ffd80 edi: f09ffb94 ebp: f05f4060
esp: f09ffb5c
Jun 1 17:08:01 nfs2 kernel: ds: 007b es: 007b ss: 0068
Jun 1 17:08:01 nfs2 kernel: Process fsstress (pid: 2284, threadinfo=f09fe000
task=f0b22d50)
Jun 1 17:08:01 nfs2 kernel: Stack: c03b1880 00000c43 c0392df2 f09ffb94
f09ffd80 f09ffd80 c01dbab1 f09ffb94
Jun 1 17:08:01 nfs2 kernel: f09ffd80 f05f4060 00000000 00000004
00000000 00000000 f05f4090 f110e004
Jun 1 17:08:01 nfs2 kernel: f05f4954 f110e004 f7a36658 f7a35380
c037bfc4 f110e000 f05f4060 f09ffd80
Jun 1 17:08:01 nfs2 kernel: Call Trace:
Jun 1 17:08:01 nfs2 kernel: [] nfs4_xdr_enc_create+0x89/0xaa
Jun 1 17:08:01 nfs2 kernel: [] rpcauth_wrap_req+0x7a/0xa6
Jun 1 17:08:01 nfs2 kernel: [] call_header+0x80/0x86
Jun 1 17:08:01 nfs2 kernel: [] nfs4_xdr_enc_create+0x0/0xaa
Jun 1 17:08:01 nfs2 kernel: [] call_encode+0xa4/0x128
Jun 1 17:08:01 nfs2 kernel: [] nfs4_xdr_enc_create+0x0/0xaa
Jun 1 17:08:01 nfs2 kernel: [] call_transmit+0x0/0xaa
Jun 1 17:08:01 nfs2 kernel: [] call_transmit+0x85/0xaa
Jun 1 17:08:01 nfs2 kernel: [] __rpc_execute+0x224/0x4d3
Jun 1 17:08:01 nfs2 kernel: [] rpcauth_bindcred+0x68/0xd0
Jun 1 17:08:01 nfs2 kernel: [] rpc_new_task+0x36/0xa7
Jun 1 17:08:01 nfs2 kernel: [] rpc_call_sync+0xc0/0xf5
Jun 1 17:08:01 nfs2 kernel: [] rpc_call_sync+0xce/0xf5
Jun 1 17:08:01 nfs2 kernel: [] pagevec_lookup+0x2e/0x38
Jun 1 17:08:01 nfs2 kernel: [] invalidate_mapping_pages+0x5e/0x10d
Jun 1 17:08:01 nfs2 kernel: [] decode_op_hdr+0x1f/0x80
Jun 1 17:08:01 nfs2 kernel: [] decode_access+0x43/0x7c
Jun 1 17:08:01 nfs2 kernel: [] decode_putfh+0x17/0x1b
Jun 1 17:08:01 nfs2 kernel: [] nfs4_proc_symlink+0xe0/0x12a
Jun 1 17:08:01 nfs2 kernel: [] nfs_symlink+0x108/0x2c3
Jun 1 17:08:01 nfs2 kernel: [] rpc_new_task+0x36/0xa7
Jun 1 17:08:01 nfs2 kernel: [] rpc_call_sync+0xce/0xf5
Jun 1 17:08:01 nfs2 kernel: [] buffered_rmqueue+0x10c/0x214
Jun 1 17:08:01 nfs2 kernel: []
rpcauth_lookup_credcache+0x22f/0x280
Jun 1 17:08:01 nfs2 kernel: [] filemap_nopage+0x1c5/0x3a2
Jun 1 17:08:01 nfs2 kernel: []
rpcauth_lookup_credcache+0x22f/0x280
Jun 1 17:08:01 nfs2 kernel: []
rpcauth_lookup_credcache+0x22f/0x280
Jun 1 17:08:01 nfs2 kernel: [] nfs4_proc_access+0xab/0x12e
Jun 1 17:08:01 nfs2 kernel: [] nfs_permission+0x233/0x29c
Jun 1 17:08:01 nfs2 kernel: [] permission+0x49/0x4b
Jun 1 17:08:01 nfs2 kernel: [] vfs_symlink+0x5f/0xb1
Jun 1 17:08:01 nfs2 kernel: [] sys_symlink+0xeb/0xfb
Jun 1 17:08:01 nfs2 kernel: [] sys_close+0x80/0xcd
Jun 1 17:08:01 nfs2 kernel: [] syscall_call+0x7/0xb
Jun 1 17:08:01 nfs2 kernel:
Jun 1 17:08:01 nfs2 kernel: Code: 0f 0b 72 02 4f 87 3a c0 eb a8 c7 44 24 04
08 00 00 00 89 3c

There is no message on the server.

Trond Myclb.:

Jun 1 17:07:45 nfs2 kernel: decode_getfattr: xdr error 10008!

This is expected. We do not yet support NFS4ERR_DELAY in the NFSv4
client.

Jun 1 17:07:55 nfs2 sshd(pam_unix)[2233]: session opened for user root by
(uid=0)
Jun 1 17:08:00 nfs2 kernel: RESERVE_SPACE(3139) failed in function
encode_create
Jun 1 17:08:00 nfs2 kernel: ------------[ cut here ]------------
Jun 1 17:08:00 nfs2 kernel: kernel BUG at fs/nfs/nfs4xdr.c:626!

This I don’t understand at all! Which line does this correspond to in
your source file? At first glance I would guess it is

    WRITEMEM(create->name->name, create->name->len);

in encode_create()?

Cheers,
Trond
v_atekor:Le Mercredi 02 Juin 2004 11:05, Trond Myklebust a écrit :

På on , 02/06/2004 klokka 01:27, skreiv Vincent ROQUETA:

Jun 1 17:07:45 nfs2 kernel: decode_getfattr: xdr error 10008!

This is expected. We do not yet support NFS4ERR_DELAY in the NFSv4
client.

Jun 1 17:07:55 nfs2 sshd(pam_unix)[2233]: session opened for user root
by (uid=0)
Jun 1 17:08:00 nfs2 kernel: RESERVE_SPACE(3139) failed in function
encode_create
Jun 1 17:08:00 nfs2 kernel: ------------[ cut here ]------------
Jun 1 17:08:00 nfs2 kernel: kernel BUG at fs/nfs/nfs4xdr.c:626!

This I don’t understand at all! Which line does this correspond to in
your source file? At first glance I would guess it is

  WRITEMEM(create->name->name, create->name->len);

in encode_create()?

/* From fs/nfs/nfs4xdr.c */

case NF4LNK:
/* Here is the “RESERVE_SPACE(3139) failed” (line 626) /
RESERVE_SPACE(4 + create->u.symlink->len); <= line 626
WRITE32(create->u.symlink->len);
/
here is the WRITEMEM */
WRITEMEM(create->u.symlink->name, create->u.symlink->len);
break;

That looks to be the correct use of WRITEMEM. What I don’t understand is why
the RESERVE_SPACE fail?

Cordialement,
/******************************************/
J’espère quel’on va pouvoir débloquer la situation pour ce soir, mais à mon avis ce en sera pas le cas :frowning:

Bon courage… :wink: