Skip to contentSkip to conversation

reiser4 for 2.6.13 is available on our website

DISCUSSION Hans Reiser 6 participants 12 messages Sep 29, 2005 – Oct 1, 2005

lore
Share & tools
lore
#0 lore raw

Reiser4 performance dropped in the -mm series due to the write throttling patch and also dropped due to a fixed bug (removing type safe lists added a bug). We don't yet know if we got quite all the performance back, we are still testing. Conservative users should wait a week or two to see if bug reports with reiser4 remain, or if indeed we did get all the bugs out of the recent cleanup patch.

Hans

BEG 2005.09.29:10.21 Thu
END 2005.09.29:14.11 Thu TOTAL=3:50
BEG 2005.09.29:14.40 Thu
CHATS: preparing reiser4 patches for 2.6.13
making sure that per-task-predictive-write-throttling-1.patch is
responsible
for reiser4 performance drop in 2.6.14-rcX-mmY
at least it is responsible for substantial part of it
END 2005.09.29:19.09 Thu TOTAL=4:29

#1 lore raw

Dnia Thu, 29 Sep 2005 19:25:54 +0200, Hans Reiser <[email protected]> napisa³:

Quoted 8 lines

Reiser4 performance dropped in the -mm series due to the write throttling patch and also dropped due to a fixed bug (removing type safe lists added a bug). We don't yet know if we got quite all the performance back, we are still testing. Conservative users should wait a week or two to see if bug reports with reiser4 remain, or if indeed we did get all the bugs out of the recent cleanup patch.

Hans

  CC      fs/proc/kmsg.o
  CC      fs/proc/proc_tty.o
  CC      fs/proc/proc_misc.o
  CC      fs/proc/kcore.o
  LD      fs/proc/proc.o
  LD      fs/proc/built-in.o
  CC      fs/ramfs/inode.o
  LD      fs/ramfs/ramfs.o
  LD      fs/ramfs/built-in.o
  CC      fs/reiser4/debug.o
In file included from fs/reiser4/lock.h:15,
                 from fs/reiser4/context.h:14,
                 from fs/reiser4/debug.c:25:

fs/reiser4/txnmgr.h: In function 'spin_atom_init':
fs/reiser4/txnmgr.h:512: error: duplicate case value fs/reiser4/txnmgr.h:512: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnh_init':
fs/reiser4/txnmgr.h:513: error: duplicate case value fs/reiser4/txnmgr.h:513: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnmgr_init':
fs/reiser4/txnmgr.h:514: error: duplicate case value fs/reiser4/txnmgr.h:514: error: previously used here In file included from fs/reiser4/context.h:14,

                 from fs/reiser4/debug.c:25:

fs/reiser4/lock.h: In function 'spin_stack_init':
fs/reiser4/lock.h:198: error: duplicate case value fs/reiser4/lock.h:198: error: previously used here In file included from fs/reiser4/znode.h:16,

                 from fs/reiser4/tree.h:15,
                 from fs/reiser4/super.h:9,
                 from fs/reiser4/debug.c:26:

fs/reiser4/jnode.h: In function 'spin_jnode_init':
fs/reiser4/jnode.h:344: error: duplicate case value fs/reiser4/jnode.h:344: error: previously used here fs/reiser4/jnode.h: In function 'spin_jload_init':
fs/reiser4/jnode.h:348: error: duplicate case value fs/reiser4/jnode.h:348: error: previously used here In file included from fs/reiser4/super.h:9,

                 from fs/reiser4/debug.c:26:

fs/reiser4/tree.h: In function 'spin_epoch_init':
fs/reiser4/tree.h:169: error: duplicate case value fs/reiser4/tree.h:169: error: previously used here In file included from fs/reiser4/debug.c:26:
fs/reiser4/super.h: In function 'spin_super_init':
fs/reiser4/super.h:379: error: duplicate case value fs/reiser4/super.h:379: error: previously used here make[2]: *** [fs/reiser4/debug.o] B³±d 1 make[1]: *** [fs/reiser4] B³±d 2
make: *** [fs] B³±d 2

amd64 gentoo box running gcc 4.0.1, 2.6.13 + reiser4-for-2.6.13-1.patch.gz

#2 lore raw

I get the same with 2.6.13.2 and gcc 3.4.4 on amd64

Quoted 63 lines · £ukasz Mierzwa

Dnia Thu, 29 Sep 2005 19:25:54 +0200, Hans Reiser <[email protected]> napisa³:

Quoted 8 lines

Reiser4 performance dropped in the -mm series due to the write throttling patch and also dropped due to a fixed bug (removing type safe lists added a bug). We don't yet know if we got quite all the performance back, we are still testing. Conservative users should wait a week or two to see if bug reports with reiser4 remain, or if indeed we did get all the bugs out of the recent cleanup patch.

Hans

   CC      fs/proc/kmsg.o
   CC      fs/proc/proc_tty.o
   CC      fs/proc/proc_misc.o
   CC      fs/proc/kcore.o
   LD      fs/proc/proc.o
   LD      fs/proc/built-in.o
   CC      fs/ramfs/inode.o
   LD      fs/ramfs/ramfs.o
   LD      fs/ramfs/built-in.o
   CC      fs/reiser4/debug.o
In file included from fs/reiser4/lock.h:15,
                  from fs/reiser4/context.h:14,
                  from fs/reiser4/debug.c:25:

fs/reiser4/txnmgr.h: In function 'spin_atom_init':
fs/reiser4/txnmgr.h:512: error: duplicate case value fs/reiser4/txnmgr.h:512: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnh_init':
fs/reiser4/txnmgr.h:513: error: duplicate case value fs/reiser4/txnmgr.h:513: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnmgr_init':
fs/reiser4/txnmgr.h:514: error: duplicate case value fs/reiser4/txnmgr.h:514: error: previously used here In file included from fs/reiser4/context.h:14,

                  from fs/reiser4/debug.c:25:

fs/reiser4/lock.h: In function 'spin_stack_init':
fs/reiser4/lock.h:198: error: duplicate case value fs/reiser4/lock.h:198: error: previously used here In file included from fs/reiser4/znode.h:16,

                  from fs/reiser4/tree.h:15,
                  from fs/reiser4/super.h:9,
                  from fs/reiser4/debug.c:26:

fs/reiser4/jnode.h: In function 'spin_jnode_init':
fs/reiser4/jnode.h:344: error: duplicate case value fs/reiser4/jnode.h:344: error: previously used here fs/reiser4/jnode.h: In function 'spin_jload_init':
fs/reiser4/jnode.h:348: error: duplicate case value fs/reiser4/jnode.h:348: error: previously used here In file included from fs/reiser4/super.h:9,

                  from fs/reiser4/debug.c:26:

fs/reiser4/tree.h: In function 'spin_epoch_init':
fs/reiser4/tree.h:169: error: duplicate case value fs/reiser4/tree.h:169: error: previously used here In file included from fs/reiser4/debug.c:26:
fs/reiser4/super.h: In function 'spin_super_init':
fs/reiser4/super.h:379: error: duplicate case value fs/reiser4/super.h:379: error: previously used here make[2]: *** [fs/reiser4/debug.o] B³±d 1 make[1]: *** [fs/reiser4] B³±d 2
make: *** [fs] B³±d 2

amd64 gentoo box running gcc 4.0.1, 2.6.13 + reiser4-for-2.6.13-1.patch.gz

Signature · 1 line
Jake Maciejewski <[email protected]>
#3 lore raw

It works on non-amd64, will fix amd64 tomorrow I hope.

#4 lore raw
Quoted 1 line · Hans Reiser

It works on non-amd64, will fix amd64 tomorrow I hope.

No, it fails on FC4/x86 here to complie as well, with the same failure mode.

#5 lore raw
Quoted 11 lines · Gregory Maxwell
Quoted 3 lines · Hans Reiser

It works on non-amd64, will fix amd64 tomorrow I hope.

No, it fails on FC4/x86 here to complie as well, with the same failure mode.

Errors from my vanilla 2.6.13.1 kernel + ReiserFS4 patch compile tonight:

fs/reiser4/super_ops.c: In function `reiser4_dirty_inode':
fs/reiser4/super_ops.c:176: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file_ops_readdir.c: In function `readdir_common':
fs/reiser4/plugin/file_ops_readdir.c:635: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file/file.c: In function `delete_object_unix_file':
fs/reiser4/plugin/file/file.c:2991: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/dir_plugin_common.c: In function `rem_entry':
fs/reiser4/plugin/dir_plugin_common.c:207: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/space/bitmap.c: In function `bnode_check_adler32':
fs/reiser4/plugin/space/bitmap.c:506: warning: long long unsigned int format, bmap_nr_t arg (arg 2)

Compile finished with no other ReiserFS related errors.

This was during a stage 1 install on Gentoo 2005.1 on a Sun Ultra II workstation (dual processor, 64-bit).

# gcc --version
gcc (GCC) 3.3.5-20050130 (Gentoo 3.3.5.20050130-r1, ssp-3.3.5.20050130-1, pie-8.7.7.1)

Are the failed compiles all on machines running GCC 4.0?

--Dan

#6 lore raw
Quoted 32 lines · Dan Oglesby
Quoted 11 lines · Gregory Maxwell
Quoted 2 lines · Hans Reiser

It works on non-amd64, will fix amd64 tomorrow I hope.

No, it fails on FC4/x86 here to complie as well, with the same failure mode.

Errors from my vanilla 2.6.13.1 kernel + ReiserFS4 patch compile tonight:

fs/reiser4/super_ops.c: In function `reiser4_dirty_inode':
fs/reiser4/super_ops.c:176: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file_ops_readdir.c: In function `readdir_common':
fs/reiser4/plugin/file_ops_readdir.c:635: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file/file.c: In function `delete_object_unix_file':
fs/reiser4/plugin/file/file.c:2991: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/dir_plugin_common.c: In function `rem_entry':
fs/reiser4/plugin/dir_plugin_common.c:207: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/space/bitmap.c: In function `bnode_check_adler32':
fs/reiser4/plugin/space/bitmap.c:506: warning: long long unsigned int format, bmap_nr_t arg (arg 2)

Compile finished with no other ReiserFS related errors.

Quoted 8 lines

This was during a stage 1 install on Gentoo 2005.1 on a Sun Ultra II workstation (dual processor, 64-bit).

# gcc --version
gcc (GCC) 3.3.5-20050130 (Gentoo 3.3.5.20050130-r1, ssp-3.3.5.20050130-1, pie-8.7.7.1)

Are the failed compiles all on machines running GCC 4.0?

--Dan

Nevermind, I just saw Jake's email (failed on gcc 3.4.4).

--Dan

#7 lore raw
Quoted 42 lines · Dan Oglesby
Quoted 40 lines · Dan Oglesby
Quoted 12 lines · Gregory Maxwell
Quoted 2 lines · Hans Reiser

It works on non-amd64, will fix amd64 tomorrow I hope.

No, it fails on FC4/x86 here to complie as well, with the same failure mode.

Errors from my vanilla 2.6.13.1 kernel + ReiserFS4 patch compile tonight:

fs/reiser4/super_ops.c: In function `reiser4_dirty_inode':
fs/reiser4/super_ops.c:176: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file_ops_readdir.c: In function `readdir_common':
fs/reiser4/plugin/file_ops_readdir.c:635: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file/file.c: In function `delete_object_unix_file':
fs/reiser4/plugin/file/file.c:2991: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/dir_plugin_common.c: In function `rem_entry':
fs/reiser4/plugin/dir_plugin_common.c:207: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/space/bitmap.c: In function `bnode_check_adler32':
fs/reiser4/plugin/space/bitmap.c:506: warning: long long unsigned int format, bmap_nr_t arg (arg 2)

Compile finished with no other ReiserFS related errors.

This was during a stage 1 install on Gentoo 2005.1 on a Sun Ultra II workstation (dual processor, 64-bit).

# gcc --version
gcc (GCC) 3.3.5-20050130 (Gentoo 3.3.5.20050130-r1, ssp-3.3.5.20050130-1, pie-8.7.7.1)

Are the failed compiles all on machines running GCC 4.0?

Quoted 8 lines
Quoted 1 line

--Dan

Nevermind, I just saw Jake's email (failed on gcc 3.4.4).

--Dan

From my Slackware 10.2 x86 system tonight:

In file included from fs/reiser4/lock.h:15,

                from fs/reiser4/context.h:14,
                from fs/reiser4/debug.c:25:

fs/reiser4/txnmgr.h: In function `spin_atom_init':
fs/reiser4/txnmgr.h:512: error: duplicate case value fs/reiser4/txnmgr.h:512: error: previously used here fs/reiser4/txnmgr.h: In function `spin_txnh_init':
fs/reiser4/txnmgr.h:513: error: duplicate case value fs/reiser4/txnmgr.h:513: error: previously used here fs/reiser4/txnmgr.h: In function `spin_txnmgr_init':
fs/reiser4/txnmgr.h:514: error: duplicate case value fs/reiser4/txnmgr.h:514: error: previously used here In file included from fs/reiser4/context.h:14,

                from fs/reiser4/debug.c:25:

fs/reiser4/lock.h: In function `spin_stack_init':
fs/reiser4/lock.h:198: error: duplicate case value fs/reiser4/lock.h:198: error: previously used here In file included from fs/reiser4/znode.h:16,

                from fs/reiser4/tree.h:15,
                from fs/reiser4/super.h:9,
                from fs/reiser4/debug.c:26:

fs/reiser4/jnode.h: In function `spin_jnode_init':
fs/reiser4/jnode.h:344: error: duplicate case value fs/reiser4/jnode.h:344: error: previously used here fs/reiser4/jnode.h: In function `spin_jload_init':
fs/reiser4/jnode.h:348: error: duplicate case value fs/reiser4/jnode.h:348: error: previously used here In file included from fs/reiser4/super.h:9,

                from fs/reiser4/debug.c:26:

fs/reiser4/tree.h: In function `spin_epoch_init':
fs/reiser4/tree.h:169: error: duplicate case value fs/reiser4/tree.h:169: error: previously used here In file included from fs/reiser4/debug.c:26:
fs/reiser4/super.h: In function `spin_super_init':
fs/reiser4/super.h:379: error: duplicate case value fs/reiser4/super.h:379: error: previously used here make[2]: *** [fs/reiser4/debug.o] Error 1 make[1]: *** [fs/reiser4] Error 2
make: *** [fs] Error 2

# gcc --version
gcc (GCC) 3.3.6

Same copy of the 2.6.13.1 kernel source and patch as the compile I did on the Gentoo machine tonight.

--Dan

#8 lore raw
Quoted 96 lines · Dan Oglesby
Quoted 43 lines · Dan Oglesby
Quoted 41 lines · Dan Oglesby
Quoted 13 lines · Gregory Maxwell
Quoted 2 lines · Hans Reiser

It works on non-amd64, will fix amd64 tomorrow I hope.

No, it fails on FC4/x86 here to complie as well, with the same failure mode.

Errors from my vanilla 2.6.13.1 kernel + ReiserFS4 patch compile tonight:

fs/reiser4/super_ops.c: In function `reiser4_dirty_inode':
fs/reiser4/super_ops.c:176: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file_ops_readdir.c: In function `readdir_common':
fs/reiser4/plugin/file_ops_readdir.c:635: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/file/file.c: In function `delete_object_unix_file':
fs/reiser4/plugin/file/file.c:2991: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/dir_plugin_common.c: In function `rem_entry':
fs/reiser4/plugin/dir_plugin_common.c:207: warning: long long unsigned int format, oid_t arg (arg 2)

fs/reiser4/plugin/space/bitmap.c: In function `bnode_check_adler32':
fs/reiser4/plugin/space/bitmap.c:506: warning: long long unsigned int format, bmap_nr_t arg (arg 2)

Compile finished with no other ReiserFS related errors.

This was during a stage 1 install on Gentoo 2005.1 on a Sun Ultra II workstation (dual processor, 64-bit).

# gcc --version
gcc (GCC) 3.3.5-20050130 (Gentoo 3.3.5.20050130-r1, ssp-3.3.5.20050130-1, pie-8.7.7.1)

Are the failed compiles all on machines running GCC 4.0?

Quoted 9 lines
Quoted 1 line

--Dan

Nevermind, I just saw Jake's email (failed on gcc 3.4.4).

--Dan

From my Slackware 10.2 x86 system tonight:

In file included from fs/reiser4/lock.h:15,

                from fs/reiser4/context.h:14,
                from fs/reiser4/debug.c:25:

fs/reiser4/txnmgr.h: In function `spin_atom_init':
fs/reiser4/txnmgr.h:512: error: duplicate case value fs/reiser4/txnmgr.h:512: error: previously used here fs/reiser4/txnmgr.h: In function `spin_txnh_init':
fs/reiser4/txnmgr.h:513: error: duplicate case value fs/reiser4/txnmgr.h:513: error: previously used here fs/reiser4/txnmgr.h: In function `spin_txnmgr_init':
fs/reiser4/txnmgr.h:514: error: duplicate case value fs/reiser4/txnmgr.h:514: error: previously used here In file included from fs/reiser4/context.h:14,

                from fs/reiser4/debug.c:25:

fs/reiser4/lock.h: In function `spin_stack_init':
fs/reiser4/lock.h:198: error: duplicate case value fs/reiser4/lock.h:198: error: previously used here In file included from fs/reiser4/znode.h:16,

                from fs/reiser4/tree.h:15,
                from fs/reiser4/super.h:9,
                from fs/reiser4/debug.c:26:

fs/reiser4/jnode.h: In function `spin_jnode_init':
fs/reiser4/jnode.h:344: error: duplicate case value fs/reiser4/jnode.h:344: error: previously used here fs/reiser4/jnode.h: In function `spin_jload_init':
fs/reiser4/jnode.h:348: error: duplicate case value fs/reiser4/jnode.h:348: error: previously used here In file included from fs/reiser4/super.h:9,

                from fs/reiser4/debug.c:26:

fs/reiser4/tree.h: In function `spin_epoch_init':
fs/reiser4/tree.h:169: error: duplicate case value fs/reiser4/tree.h:169: error: previously used here In file included from fs/reiser4/debug.c:26:
fs/reiser4/super.h: In function `spin_super_init':
fs/reiser4/super.h:379: error: duplicate case value fs/reiser4/super.h:379: error: previously used here make[2]: *** [fs/reiser4/debug.o] Error 1 make[1]: *** [fs/reiser4] Error 2
make: *** [fs] Error 2

Quoted 6 lines
# gcc --version
gcc (GCC) 3.3.6

Same copy of the 2.6.13.1 kernel source and patch as the compile I did on the Gentoo machine tonight.

--Dan

Last update for tonight/this morning...

On a Slackware 10.1 machine (32bit/x86), with gcc 3.3.4 there were no warnings or errors related to ReiserFS4 (got a couple with some deprecated bits of serial port code, nothing from filesystem code).

I'll definately get some ReiserFS4 filesystems online this weekend, and see how hard I can push this FS on my various machines.

--Dan

#9 lore raw
In reply to Jake Maciejewski
Quoted 2 lines · Jake Maciejewski

I get the same with 2.6.13.2 and gcc 3.4.4 on amd64

can you try this patch?
-----------------------------------

fs/reiser4/spin_macros.h +0 -3
@@ -82,8 +82,6 @@ typedef struct reiser4_rw_data {
 static inline void spin_ ## NAME ## _init(TYPE *x)				\
 {										\
 	__ODCA("nikita-2987", x != NULL);					\
-	cassert(sizeof(x->FIELD) != 0);						\
-	memset(& x->FIELD, 0, sizeof x->FIELD);					\
 	spin_lock_init(& x->FIELD.lock);					\
 }										\
 										\
@@ -236,7 +234,6 @@ typedef struct { int foo; } NAME ## _spi
 static inline void rw_ ## NAME ## _init(TYPE *x)				\
 {										\
 	__ODCA("nikita-2988", x != NULL);					\
-	memset(& x->FIELD, 0, sizeof x->FIELD);					\
 	rwlock_init(& x->FIELD.lock);						\
 }										\
 										\

-----------------------------------

Quoted 65 lines · £ukasz Mierzwa
Quoted 64 lines

Dnia Thu, 29 Sep 2005 19:25:54 +0200, Hans Reiser <[email protected]>

napisa³:

Quoted 8 lines

Reiser4 performance dropped in the -mm series due to the write throttling patch and also dropped due to a fixed bug (removing type safe lists added a bug). We don't yet know if we got quite all the performance back, we are still testing. Conservative users should wait a week or two to see if bug reports with reiser4 remain, or if indeed we did get all the bugs out of the recent cleanup patch.

Hans

   CC      fs/proc/kmsg.o
   CC      fs/proc/proc_tty.o
   CC      fs/proc/proc_misc.o
   CC      fs/proc/kcore.o
   LD      fs/proc/proc.o
   LD      fs/proc/built-in.o
   CC      fs/ramfs/inode.o
   LD      fs/ramfs/ramfs.o
   LD      fs/ramfs/built-in.o
   CC      fs/reiser4/debug.o
In file included from fs/reiser4/lock.h:15,
                  from fs/reiser4/context.h:14,
                  from fs/reiser4/debug.c:25:

fs/reiser4/txnmgr.h: In function 'spin_atom_init':
fs/reiser4/txnmgr.h:512: error: duplicate case value fs/reiser4/txnmgr.h:512: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnh_init':
fs/reiser4/txnmgr.h:513: error: duplicate case value fs/reiser4/txnmgr.h:513: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnmgr_init':
fs/reiser4/txnmgr.h:514: error: duplicate case value fs/reiser4/txnmgr.h:514: error: previously used here In file included from fs/reiser4/context.h:14,

                  from fs/reiser4/debug.c:25:

fs/reiser4/lock.h: In function 'spin_stack_init':
fs/reiser4/lock.h:198: error: duplicate case value fs/reiser4/lock.h:198: error: previously used here In file included from fs/reiser4/znode.h:16,

                  from fs/reiser4/tree.h:15,
                  from fs/reiser4/super.h:9,
                  from fs/reiser4/debug.c:26:

fs/reiser4/jnode.h: In function 'spin_jnode_init':
fs/reiser4/jnode.h:344: error: duplicate case value fs/reiser4/jnode.h:344: error: previously used here fs/reiser4/jnode.h: In function 'spin_jload_init':
fs/reiser4/jnode.h:348: error: duplicate case value fs/reiser4/jnode.h:348: error: previously used here In file included from fs/reiser4/super.h:9,

                  from fs/reiser4/debug.c:26:

fs/reiser4/tree.h: In function 'spin_epoch_init':
fs/reiser4/tree.h:169: error: duplicate case value fs/reiser4/tree.h:169: error: previously used here In file included from fs/reiser4/debug.c:26:
fs/reiser4/super.h: In function 'spin_super_init':
fs/reiser4/super.h:379: error: duplicate case value fs/reiser4/super.h:379: error: previously used here make[2]: *** [fs/reiser4/debug.o] B³±d 1 make[1]: *** [fs/reiser4] B³±d 2
make: *** [fs] B³±d 2

amd64 gentoo box running gcc 4.0.1, 2.6.13 + reiser4-for-2.6.13-1.patch.gz

#10 lore raw

Dnia Fri, 30 Sep 2005 10:31:01 +0200, Alexander Zarochentsev <[email protected]> napisa³:

Quoted 17 lines · Jake Maciejewski
Quoted 2 lines

I get the same with 2.6.13.2 and gcc 3.4.4 on amd64

can you try this patch?
-----------------------------------

fs/reiser4/spin_macros.h +0 -2
@@ -82,8 +82,6 @@ typedef struct reiser4_rw_data {
 static inline void spin_ ## NAME ## _init(TYPE *x)				\
 {										\
 	__ODCA("nikita-2987", x != NULL);					\
-	cassert(sizeof(x->FIELD) != 0);						\
-	memset(& x->FIELD, 0, sizeof x->FIELD);					\
 	spin_lock_init(& x->FIELD.lock);					\
 }										\
 										\

fs/reiser4/spin_macros.h typedef struct { int foo; } NAME ## _spi

Quoted 8 lines
fs/reiser4/spin_macros.h +0 -2
@@ -236,7 +234,6 @@ typedef struct { int foo; } NAME ## _spi
 static inline void rw_ ## NAME ## _init(TYPE *x)				\
 {										\
 	__ODCA("nikita-2988", x != NULL);					\
-	memset(& x->FIELD, 0, sizeof x->FIELD);					\
 	rwlock_init(& x->FIELD.lock);						\
 }										\
 										\
-----------------------------------

It helped (cut and paste didn't worked for me as the resulting patch didn't applied but I just deleted those 3 lines by hand). Kernel did compile and now I'm writing this running 2.6.13 with reiser4 as /. I hope I won't see any oopses during shutdown as I sow with 2.6.13-mm{1,2} based kernels (archck).

#11 lore raw
In reply to Alexander Zarochentsev

It compiles and boots, but I can't stress test because two of my NICs don't work in 2.6.13.

I don't know which patch(es) fixed my NIC problem, but I recall 2.6.14-rc1 working. When I tried the 2.6.13-1 patch against rc1, I got a failed hunk in lib/radix-tree.c which I attempted to fix manually. It compiled, but I got a panic starting init.

Quoted 90 lines · Alexander Zarochentsev
Quoted 2 lines · Jake Maciejewski

I get the same with 2.6.13.2 and gcc 3.4.4 on amd64

can you try this patch?
-----------------------------------

fs/reiser4/spin_macros.h +0 -4
@@ -82,8 +82,6 @@ typedef struct reiser4_rw_data {
 static inline void spin_ ## NAME ## _init(TYPE *x)				\
 {										\
 	__ODCA("nikita-2987", x != NULL);					\
-	cassert(sizeof(x->FIELD) != 0);						\
-	memset(& x->FIELD, 0, sizeof x->FIELD);					\
 	spin_lock_init(& x->FIELD.lock);					\
 }										\
 										\
@@ -236,7 +234,6 @@ typedef struct { int foo; } NAME ## _spi
 static inline void rw_ ## NAME ## _init(TYPE *x)				\
 {										\
 	__ODCA("nikita-2988", x != NULL);					\
-	memset(& x->FIELD, 0, sizeof x->FIELD);					\
 	rwlock_init(& x->FIELD.lock);						\
 }										\
 										\
-----------------------------------
Quoted 65 lines · £ukasz Mierzwa
Quoted 64 lines

Dnia Thu, 29 Sep 2005 19:25:54 +0200, Hans Reiser <[email protected]>

napisa³:

Quoted 8 lines

Reiser4 performance dropped in the -mm series due to the write throttling patch and also dropped due to a fixed bug (removing type safe lists added a bug). We don't yet know if we got quite all the performance back, we are still testing. Conservative users should wait a week or two to see if bug reports with reiser4 remain, or if indeed we did get all the bugs out of the recent cleanup patch.

Hans

   CC      fs/proc/kmsg.o
   CC      fs/proc/proc_tty.o
   CC      fs/proc/proc_misc.o
   CC      fs/proc/kcore.o
   LD      fs/proc/proc.o
   LD      fs/proc/built-in.o
   CC      fs/ramfs/inode.o
   LD      fs/ramfs/ramfs.o
   LD      fs/ramfs/built-in.o
   CC      fs/reiser4/debug.o
In file included from fs/reiser4/lock.h:15,
                  from fs/reiser4/context.h:14,
                  from fs/reiser4/debug.c:25:

fs/reiser4/txnmgr.h: In function 'spin_atom_init':
fs/reiser4/txnmgr.h:512: error: duplicate case value fs/reiser4/txnmgr.h:512: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnh_init':
fs/reiser4/txnmgr.h:513: error: duplicate case value fs/reiser4/txnmgr.h:513: error: previously used here fs/reiser4/txnmgr.h: In function 'spin_txnmgr_init':
fs/reiser4/txnmgr.h:514: error: duplicate case value fs/reiser4/txnmgr.h:514: error: previously used here In file included from fs/reiser4/context.h:14,

                  from fs/reiser4/debug.c:25:

fs/reiser4/lock.h: In function 'spin_stack_init':
fs/reiser4/lock.h:198: error: duplicate case value fs/reiser4/lock.h:198: error: previously used here In file included from fs/reiser4/znode.h:16,

                  from fs/reiser4/tree.h:15,
                  from fs/reiser4/super.h:9,
                  from fs/reiser4/debug.c:26:

fs/reiser4/jnode.h: In function 'spin_jnode_init':
fs/reiser4/jnode.h:344: error: duplicate case value fs/reiser4/jnode.h:344: error: previously used here fs/reiser4/jnode.h: In function 'spin_jload_init':
fs/reiser4/jnode.h:348: error: duplicate case value fs/reiser4/jnode.h:348: error: previously used here In file included from fs/reiser4/super.h:9,

                  from fs/reiser4/debug.c:26:

fs/reiser4/tree.h: In function 'spin_epoch_init':
fs/reiser4/tree.h:169: error: duplicate case value fs/reiser4/tree.h:169: error: previously used here In file included from fs/reiser4/debug.c:26:
fs/reiser4/super.h: In function 'spin_super_init':
fs/reiser4/super.h:379: error: duplicate case value fs/reiser4/super.h:379: error: previously used here make[2]: *** [fs/reiser4/debug.o] B³±d 1 make[1]: *** [fs/reiser4] B³±d 2
make: *** [fs] B³±d 2

amd64 gentoo box running gcc 4.0.1, 2.6.13 + reiser4-for-2.6.13-1.patch.gz

Signature · 1 line
Jake Maciejewski <[email protected]>