Quota support for UID >65534

Quota support for UID >65534

[Zeljko Brajdic]

Hello,

I've applied quota patches for vanilla kernel-2.4.18 found at
ftp://ftp.suse.com/pub/people/mason/patches/reiserfs/quota-2.4/2.4.18/
Everything works perfectly except for checking quota with command `quota
-v username' if username have UID >65534. Interesting is that `repquota'
doesn't have that problem and showing.

This is output of `quota':

root@led 12:02 (/usr/src/linux) > quota -v bureq
quota: Error while getting quota from /dev/sda4 for 200000: Invalid argument
Disk quotas for user bureq (uid 200000): none


Strace output is (last few lines):

stat64("/dev/sda4", {st_mode=S_IFBLK|0660, st_rdev=makedev(8, 4), ...}) = 0
getuid32()                              = 0
quotactl(0xd00 /* Q_??? */|USRQUOTA, "/dev/sda4", 0, {0, 0, 2, 0, 0, 2538116, 0, 0}) = 0
quotactl(0x900 /* Q_??? */|USRQUOTA, "/dev/sda4", 0, {604800, 604800, 0, 2476, 0, 2475, 3221224236, 134536589}) = 0
geteuid32()                             = 0
quotactl(0xd00 /* Q_??? */|USRQUOTA, "/dev/sda4", 200000, {1073780232, 1073821628, 1073823040, 1, 0, 0, 0, 0}) = -1 EINVAL (Invalid argument)
write(2, "quota: ", 7quota: )                  = 7
write(2, "Error while getting quota from /"..., 70Error while getting quota from /dev/sda4 for 200000: Invalid argument
) = 70
fstat64(1, {st_mode=S_IFCHR|0600, st_rdev=makedev(136, 0), ...}) = 0
old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40014000
write(1, "Disk quotas for user bureq (uid "..., 46Disk quotas for user bureq (uid 200000): none
) = 46
munmap(0x40014000, 4096)                = 0
_exit(0)                                = ?


-- 
v            ,   v  v
Zeljko Brajdic - Zorz

Re: Quota support for UID >65534

[Chris Mason]

On Mon, 2002-07-15 at 06:11, Zeljko Brajdic wrote:
> Hello,
> 
> I've applied quota patches for vanilla kernel-2.4.18 found at
> ftp://ftp.suse.com/pub/people/mason/patches/reiserfs/quota-2.4/2.4.18/
> Everything works perfectly except for checking quota with command `quota
> -v username' if username have UID >65534. Interesting is that `repquota'
> doesn't have that problem and showing.
> 
> This is output of `quota':
> 
> root@led 12:02 (/usr/src/linux) > quota -v bureq
> quota: Error while getting quota from /dev/sda4 for 200000: Invalid argument
> Disk quotas for user bureq (uid 200000): none

Which disk format are you using?  You can check this with debugreiserfs
/dev/xxxxx

-chris

Re: Quota support for UID >65534

[Zeljko Brajdic]

On Mon, 2002-07-15 at 13:09, Chris Mason wrote:
> On Mon, 2002-07-15 at 06:11, Zeljko Brajdic wrote:
> > Hello,
> > 
> > I've applied quota patches for vanilla kernel-2.4.18 found at
> > ftp://ftp.suse.com/pub/people/mason/patches/reiserfs/quota-2.4/2.4.18/
> > Everything works perfectly except for checking quota with command `quota
> > -v username' if username have UID >65534. Interesting is that `repquota'
> > doesn't have that problem and showing.
> > 
> > This is output of `quota':
> > 
> > root@led 12:02 (/usr/src/linux) > quota -v bureq
> > quota: Error while getting quota from /dev/sda4 for 200000: Invalid argument
> > Disk quotas for user bureq (uid 200000): none
> 
> Which disk format are you using?  You can check this with debugreiserfs
> /dev/xxxxx
> 
root@led 13:12 (/usr/src/linux) > debugreiserfs /dev/sda4

<-------------debugreiserfs, 2002------------->
reiserfsprogs 3.x.1b


Filesystem state: consistency is not checked after last mounting

Reiserfs super block in block 16 on 0x804 of format 3.6 with standard journal


We have found "error". :) In fs/dquot.c at line 2057 exists if
statement:

        if (id & ~0xFFFF)
                goto out;

Is this feature or bug?! :) This have been found in vanilla kernel-2.4.18
-- 
v            ,   v  v
Zeljko Brajdic - Zorz

Re: Quota support for UID >65534

[Chris Mason]

On Tue, 2002-07-16 at 04:14, Zeljko Brajdic wrote:
> On Mon, 2002-07-15 at 13:09, Chris Mason wrote:
> > On Mon, 2002-07-15 at 06:11, Zeljko Brajdic wrote:
> > > Hello,
> > > 
> > > I've applied quota patches for vanilla kernel-2.4.18 found at
> > > ftp://ftp.suse.com/pub/people/mason/patches/reiserfs/quota-2.4/2.4.18/
> > > Everything works perfectly except for checking quota with command `quota
> > > -v username' if username have UID >65534. Interesting is that `repquota'
> > > doesn't have that problem and showing.
> > > 
> > > This is output of `quota':
> > > 
> > > root@led 12:02 (/usr/src/linux) > quota -v bureq
> > > quota: Error while getting quota from /dev/sda4 for 200000: Invalid argument
> > > Disk quotas for user bureq (uid 200000): none
> > 
> > Which disk format are you using?  You can check this with debugreiserfs
> > /dev/xxxxx
> > 
> root@led 13:12 (/usr/src/linux) > debugreiserfs /dev/sda4
> 
> <-------------debugreiserfs, 2002------------->
> reiserfsprogs 3.x.1b
> 
> 
> Filesystem state: consistency is not checked after last mounting
> 
> Reiserfs super block in block 16 on 0x804 of format 3.6 with standard journal

This means your reiserfs disk can support the high uids.

> 
> 
> We have found "error". :) In fs/dquot.c at line 2057 exists if
> statement:
> 
>         if (id & ~0xFFFF)
>                 goto out;
> 
> Is this feature or bug?! :) This have been found in vanilla kernel-2.4.18

This check is not in a quota patched kernel.  sys_quotactl isn't in
fs/dquot.c with the quota patches applied.

Did you apply all of the patches in my quota directory?  I think so,
since the kernel wouldn't compile otherwise ;-)

quota -v high_uid_user works fine for me, which version of the quota
utils do you have installed?

-chris

Re: Quota support for UID >65534

[Zorz]

On Tue, 2002-07-16 at 14:25, Chris Mason wrote:
> On Tue, 2002-07-16 at 04:14, Zeljko Brajdic wrote:
> > On Mon, 2002-07-15 at 13:09, Chris Mason wrote:
> > > On Mon, 2002-07-15 at 06:11, Zeljko Brajdic wrote:
> > > > Hello,
> > > > 
> > > > I've applied quota patches for vanilla kernel-2.4.18 found at
> > > > ftp://ftp.suse.com/pub/people/mason/patches/reiserfs/quota-2.4/2.4.18/
> > > > Everything works perfectly except for checking quota with command `quota
> > > > -v username' if username have UID >65534. Interesting is that `repquota'
> > > > doesn't have that problem and showing.
> > > > 
> > > > This is output of `quota':
> > > > 
> > > > root@led 12:02 (/usr/src/linux) > quota -v bureq
> > > > quota: Error while getting quota from /dev/sda4 for 200000: Invalid argument
> > > > Disk quotas for user bureq (uid 200000): none
> > > 
> > > Which disk format are you using?  You can check this with debugreiserfs
> > > /dev/xxxxx
> > > 
> > root@led 13:12 (/usr/src/linux) > debugreiserfs /dev/sda4
> > 
> > <-------------debugreiserfs, 2002------------->
> > reiserfsprogs 3.x.1b
> > 
> > 
> > Filesystem state: consistency is not checked after last mounting
> > 
> > Reiserfs super block in block 16 on 0x804 of format 3.6 with standard journal
> 
> This means your reiserfs disk can support the high uids.
> 
> > 
> > 
> > We have found "error". :) In fs/dquot.c at line 2057 exists if
> > statement:
> > 
> >         if (id & ~0xFFFF)
> >                 goto out;
> > 
> > Is this feature or bug?! :) This have been found in vanilla kernel-2.4.18
> 
> This check is not in a quota patched kernel.  sys_quotactl isn't in
> fs/dquot.c with the quota patches applied.
> 
> Did you apply all of the patches in my quota directory?  I think so,
> since the kernel wouldn't compile otherwise ;-)
> 
Yes, i've applied: 50_quota-patch-2.4.15-1-2.4.16, dquot_deadlock.diff, 
kinoded-8.diff, nesting-10.diff,  reiserfs-quota-20.diff

> quota -v high_uid_user works fine for me, which version of the quota
> utils do you have installed?
> 
root@led 15:03 (~) > quota -V
Quota utilities version 3.06.
-- 
v            ,   v  v
Zeljko Brajdic - Zorz

Re: Quota support for UID >65534

[Philippe Gramoullé]

On 16 Jul 2002 10:14:08 +0200
Zeljko Brajdic <zorz@iskon.hr> wrote:


Following Honza recommandations , we simply stripped out the 
"if (id & ~0xFFFF)" line by hand and we now use quota ith 32bits UIDs without
any problem :o)

It's not a bug , more likely a test that's been there for long at the time
of 32bit UIDs were not ready , it think.

Thanks,

Philippe.

  |  We have found "error". :) In fs/dquot.c at line 2057 exists if
  |  statement:
  |  
  |          if (id & ~0xFFFF)
  |                  goto out;
  |  
  |  Is this feature or bug?! :) This have been found in vanilla kernel-2.4.18
  |  -- 
  |  v            ,   v  v
  |  Zeljko Brajdic - Zorz
  |  
  |