Edward Shishkin wrote:
> there are metas-patches on our ftp-server, although it is > not fresh. I'll advance it on the next week, ok? > > Edward. Attached is a patch for 2.6.15 that I threw together from the 2.5.14-rc5-mm1 patch on the ftp server. Please check if it's ok. Have a few problems/observations: 1. I get an oops if I try `cat items`. Reproduced with 2.6.14-rc5-mm1 andit's patch on the namesys ftp server, so it's been there for a while.$ cd test/.... $ cat items Unable to handle kernel NULL pointer dereference at virtual address 00000000 printing eip: f8f32083 *pde = 00000000 Oops: 0000 [#1] PREEMPTModules linked in: reiser4 i2c_i801 i2c_core intel_agp agpgart ipw2200ieee80211 ieee80211_crypt CPU: 0 EIP: 0060:[<f8f32083>] Not tainted VLI EFLAGS: 00010286 (2.6.15.1 <http://2.6.15.1>) EIP is at get_current_lock_stack+0x13/0x80 [reiser4] eax: 00000000 ebx: cadd9000 ecx: 00000000 edx: 00000000 esi: d6a357fc edi: ee151c10 ebp: ee151bf8 esp: cadd9db8 ds: 007b es: 007b ss: 0068 Process cat (pid: 17667, threadinfo=cadd9000 task=ddfa4a50)Stack: c0514a14 fffffff5 de4e9005 de4e9005 00000000 cadd9e0c c017511fea5763c0cadd9e0c ee151bf8 f8f705d9 cadd9e14 00000000 00000000 c01443cbd69463a000000101 dbc1aebc cadd9f38 dfff4d40 00000000 c017eb55 ea5763c000000000 Call Trace: [<c017511f>] __link_path_walk+0xbbf/0xf50 [<f8f705d9>] object_lookup+0x79/0x380 [reiser4] [<c01443cb>] do_generic_mapping_read+0x4bb/0x570 [<c017eb55>] dput+0x55/0x280 [<c0147be4>] bad_range+0x34/0x50 [<c01487b2>] buffered_rmqueue+0x1e2/0x240 [<f8f9ff07>] key_by_inode_and_offset_common+0x17/0x230 [reiser4] [<f8f9b060>] permission_common+0x0/0x10 [reiser4] [<f9014970>] get_inode_host+0x10/0xd0 [reiser4] [<f9016791>] items_start+0xe1/0x230 [reiser4] [<c0188f3b>] seq_read+0x16b/0x2f0 [<c016536c>] vfs_read+0x19c/0x1b0 [<c01656b1>] sys_read+0x51/0x80 [<c010316b>] sysenter_past_esp+0x54/0x75 Code: af c3 8b 74 24 0c 8b 5c 24 08 83 c4 10 c3 90 90 90 90 90 90 90 90 90 90 53 83 ec 24 bb 00 f0 ff ff 21 e3 8b 03 8b 80 ac 04 00 00 <81> 38 0b 5d 1b 4b 75 08 83 c4 24 83 c0 04 5b c3 e8 68 26 fe ff 2. I also get an oops using `cat readir` in a directory. Reproduced with 2.6.14-rc5-mm1 and it's patch on the ftp server. $ cd testdir/.... $ cat readdir Unable to handle kernel NULL pointer dereference at virtual address 00000000 printing eip: f8f32083 *pde = 00000000 Oops: 0000 [#1] PREEMPT Modules linked in: reiser4 i2c_i801 i2c_core intel_agp agpgart ipw2200 ieee80211 ieee80211_crypt CPU: 0 EIP: 0060:[<f8f32083>] Not tainted VLI EFLAGS: 00010286 (2.6.15.1 <http://2.6.15.1>) EIP is at get_current_lock_stack+0x13/0x80 [reiser4] eax: 00000000 ebx: ef00f000 ecx: 00000000 edx: 00000000 esi: f0ee2c7c edi: ed88de10 ebp: ed88ddf8 esp: ef00fdac ds: 007b es: 007b ss: 0068 Process cat (pid: 1649, threadinfo=ef00f000 task=ef5cbad0)Stack: f7f8a007 c017eb55 ef420998 c0514a14 fffffff5 f7f8a007 f7f8a00700000000ef00fe0c ed88ddf8 f8f705d9 ef00fe08 ef00fe04 ef00f000 c17f59800000000000000000 c01443cb f73f5b48 00000101 f0eed37c ef00ff38 dfff4d4000000000 Call Trace: [<c017eb55>] dput+0x55/0x280 [<f8f705d9>] object_lookup+0x79/0x380 [reiser4] [<c01443cb>] do_generic_mapping_read+0x4bb/0x570 [<c0175532>] link_path_walk+0x82/0xf0 [<c0147be4>] bad_range+0x34/0x50 [<c01487b2>] buffered_rmqueue+0x1e2/0x240 [<f8fcb335>] build_entry_key_hashed+0x75/0x2a0 [reiser4] [<f8f9b060>] permission_common+0x0/0x10 [reiser4] [<f9014970>] get_inode_host+0x10/0xd0 [reiser4] [<f90159cb>] readdir_start+0x10b/0x200 [reiser4] [<c0188f3b>] seq_read+0x16b/0x2f0 [<c016536c>] vfs_read+0x19c/0x1b0 [<c01656b1>] sys_read+0x51/0x80 [<c010316b>] sysenter_past_esp+0x54/0x75 Code: af c3 8b 74 24 0c 8b 5c 24 08 83 c4 10 c3 90 90 90 90 90 90 90 90 90 90 53 83 ec 24 bb 00 f0 ff ff 21 e3 8b 03 8b 80 ac 04 00 00 <81> 38 0b 5d 1b 4b 75 08 83 c4 24 83 c0 04 5b c3 e8 68 26 fe ff
Thanks for the report, I put fixed patches (including the one against 2.6.15) on our ftp-server.
3. At http://www.namesys.com/v4/pseudo.html<http://www.namesys.com/v4/pseudo.html> there is a _pagecache_ pseudofile that seems to be missing. There is also a pseudo file called _new_ that isn't described there. What is it for?
This is an alternative interface to create regular files with specified names (currently within directory): try echo -e "foo\0" > some_dir/..../new
4. I couldn't reproduce the bash-crashing bug described at http://pvh.ca/trac/wiki/Reiser4Bugs Has it been fixed?
Perhaps it got fixed when migrating to the new code for reiser4/vfs interface (Peter used the old one). Edward.
