htree directory indexing 2.4.18-2 BUG with highmem and also high i/o

htree directory indexing 2.4.18-2 BUG with highmem and also high i/o

[Marc-Christian Petersen]

Hi Daniel,

I've found a bug with htree directory indexing patch and
highmem enabled (64GB). This is with 2.4.18 and htree patch
2.4.18-2. Oops appears if accessing an ext2 partition with ls
or doing "who/w" in the directory of the ext2 partition.

Without the patch all works fine.


ksymoops 2.4.5 on i686 2.4.18.  Options used
     -V (default)
     -k /proc/ksyms (default)
     -l /proc/modules (default)
     -o /lib/modules/2.4.18/ (default)
     -m /boot/System.map-2.4.18 (default)

Warning: You did not tell me where to find symbol information.  I will
assume that the log matches the kernel and modules that are running
right now and I'll use the default options above for symbol resolution.
If the current kernel and/or modules do not match the log, you can get
more accurate output by telling me the kernel version and where to find
map, modules, ksyms etc.  ksymoops -h explains the options.

Jul  9 22:47:56 ml570 kernel: Unable to handle kernel NULL pointer dereference 
at virtual address 00000006
Jul  9 22:47:56 ml570 kernel: c019e190
Jul  9 22:47:56 ml570 kernel: *pde = 2a535001
Jul  9 22:47:56 ml570 kernel: Oops: 0000
Jul  9 22:47:56 ml570 kernel: CPU:    1
Jul  9 22:47:56 ml570 kernel: EIP:    0010:[ext2_find_entry+564/836]    Not 
tainted
Jul  9 22:47:56 ml570 kernel: EFLAGS: 00010293
Jul  9 22:47:56 ml570 kernel: eax: 00000400   ebx: 00000000   ecx: d3129900   
edx: 000003f8
Jul  9 22:47:56 ml570 kernel: esi: 00000000   edi: ea080900   ebp: d3133ea4   
esp: d3133e5c
Jul  9 22:47:56 ml570 kernel: ds: 0018   es: 0018   ss: 0018
Jul  9 22:47:56 ml570 kernel: Process w (pid: 371, stackpage=d3133000)
Jul  9 22:47:56 ml570 kernel: Stack: ea080900 00000000 ea080900 d3129900 
c0139535 d3132000 00000000 00000400 
Jul  9 22:47:56 ml570 kernel:        0000000a 00000003 ea080960 00000400 
d31cc500 c01336da e0656b80 001c8000 
Jul  9 22:47:56 ml570 kernel:        00000000 00000246 e925e100 c019e2eb 
e925e100 ea080900 d3133ec0 ea080900 
Jul  9 22:47:56 ml570 kernel: Call Trace: [filemap_nopage+241/524] 
[do_no_page+182/940] [ext2_inode_by_name+27/128] 
Warning (Oops_read): Code line not seen, dumping what data is available

>>ecx; d3129900 <[iptable_nat].bss.end+1a4a6d/8ba16d>
>>edi; ea080900 <END_OF_CODE+1648fb41/????>
>>ebp; d3133ea4 <[iptable_nat].bss.end+1af011/8ba16d>
>>esp; d3133e5c <[iptable_nat].bss.end+1aefc9/8ba16d>

Jul  9 22:47:56 ml570 kernel: Code: 0f b6 43 06 39 45 dc 75 1a 83 3b 00 74 15 
8b 75 e0 8b 4d dc 
Using defaults from ksymoops -t elf32-i386 -a i386


Code;  00000000 Before first symbol
00000000 <_EIP>:
Code;  00000000 Before first symbol
   0:   0f b6 43 06               movzbl 0x6(%ebx),%eax
Code;  00000004 Before first symbol
   4:   39 45 dc                  cmp    %eax,0xffffffdc(%ebp)
Code;  00000007 Before first symbol
   7:   75 1a                     jne    23 <_EIP+0x23> 00000023 Before first 
symbol
Code;  00000009 Before first symbol
   9:   83 3b 00                  cmpl   $0x0,(%ebx)
Code;  0000000c Before first symbol
   c:   74 15                     je     23 <_EIP+0x23> 00000023 Before first 
symbol
Code;  0000000e Before first symbol
   e:   8b 75 e0                  mov    0xffffffe0(%ebp),%esi
Code;  00000011 Before first symbol
  11:   8b 4d dc                  mov    0xffffffdc(%ebp),%ecx

-- 
Kind regards,
        Marc-Christian Petersen

http://sourceforge.net/projects/wolk

PGP/GnuPG Key: 1024D/408B2D54947750EC
Fingerprint: 8602 69E0 A9C2 A509 8661 2B0B 408B 2D54 9477 50EC
Key available at www.keyserver.net. Encrypted e-mail preferred.

Re: htree directory indexing 2.4.18-2 BUG with highmem and also high i/o

[Andreas Dilger]

On Jul 09, 2002  23:33 +0200, Marc-Christian Petersen wrote:
> Hi Daniel,
> 
> I've found a bug with htree directory indexing patch and
> highmem enabled (64GB). This is with 2.4.18 and htree patch
> 2.4.18-2. Oops appears if accessing an ext2 partition with ls
> or doing "who/w" in the directory of the ext2 partition.

The ext2 htree patch probably needs to add a "kmap()" and "kunmap()"
in the function that reads a page and scans the directory for the
name it is looking for.  I can't be any more specific than this
right now since I only have the ext3 version of this patch, and it
does not have page-cache based directories (it is still using the
buffer cache).

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/

Re: htree directory indexing 2.4.18-2 BUG with highmem and also high i/o

[Daniel Phillips]

On Wednesday 10 July 2002 23:01, Andreas Dilger wrote:
> On Jul 09, 2002  23:33 +0200, Marc-Christian Petersen wrote:
> > Hi Daniel,
> > 
> > I've found a bug with htree directory indexing patch and
> > highmem enabled (64GB). This is with 2.4.18 and htree patch
> > 2.4.18-2. Oops appears if accessing an ext2 partition with ls
> > or doing "who/w" in the directory of the ext2 partition.
> 
> The ext2 htree patch probably needs to add a "kmap()" and "kunmap()"
> in the function that reads a page and scans the directory for the
> name it is looking for.  I can't be any more specific than this
> right now since I only have the ext3 version of this patch, and it
> does not have page-cache based directories (it is still using the
> buffer cache).

Hi Andreas,

Yes, kmap is utterly not-there for the Ext2 version of the patch.
This is one of those uninteresting but altogether essential chores
that just needs doing.  Sigh.

See you tomorrow.

-- 
Daniel

Re: htree directory indexing 2.4.18-2 BUG with highmem and also high i/o

[Marc-Christian Petersen]

On Thursday 11 July 2002 00:46, Daniel Phillips wrote:

Hi Daniel, Hi Andreas,

> On Wednesday 10 July 2002 23:01, Andreas Dilger wrote:
> > On Jul 09, 2002  23:33 +0200, Marc-Christian Petersen wrote:
> > > Hi Daniel,
> > >
> > > I've found a bug with htree directory indexing patch and
> > > highmem enabled (64GB). This is with 2.4.18 and htree patch
> > > 2.4.18-2. Oops appears if accessing an ext2 partition with ls
> > > or doing "who/w" in the directory of the ext2 partition.
> >
> > The ext2 htree patch probably needs to add a "kmap()" and "kunmap()"
> > in the function that reads a page and scans the directory for the
> > name it is looking for.  I can't be any more specific than this
> > right now since I only have the ext3 version of this patch, and it
> > does not have page-cache based directories (it is still using the
> > buffer cache).

Say, where can i find those patch for ext3? I've searched a long time and 
never found it?! :|


-- 
Kind regards
        Marc-Christian Petersen

http://sourceforge.net/projects/wolk

PGP/GnuPG Key: 1024D/408B2D54947750EC
Fingerprint: 8602 69E0 A9C2 A509 8661 2B0B 408B 2D54 9477 50EC
Key available at www.keyserver.net. Encrypted e-mail preferred.

Re: htree directory indexing 2.4.18-2 BUG with highmem and also high i/o

[Andreas Dilger]

On Jul 11, 2002  01:03 +0200, Marc-Christian Petersen wrote:
> > On Wednesday 10 July 2002 23:01, Andreas Dilger wrote:
> > > The ext2 htree patch probably needs to add a "kmap()" and "kunmap()"
> > > in the function that reads a page and scans the directory for the
> > > name it is looking for.  I can't be any more specific than this
> > > right now since I only have the ext3 version of this patch, and it
> > > does not have page-cache based directories (it is still using the
> > > buffer cache).
> 
> Say, where can i find those patch for ext3? I've searched a long time and 
> never found it?! :|

The most up-to-date version is in the ext3 CVS tree at sourceforge:
http://sourceforge.net/projects/gkernel/

the ext3 package therein has "features-branch" on which the ext3 htree
patch has already been applied (along with other ext3 fixes).  If you
only want the ext3 htree patch, you can also get it from the "lustre"
sourceforge project at "lustre/extN/htree-ext3-2.4.18.diff".

Cheers, Andreas
--
Andreas Dilger
http://www-mddsp.enel.ucalgary.ca/People/adilger/
http://sourceforge.net/projects/ext2resize/