Re: Weird /proc/meminfo output on 2.4.13-ac5

Re: Weird /proc/meminfo output on 2.4.13-ac5

[Petr Vandrovec]

On  2 Nov 01 at 14:10, Andreas Franck wrote:
> $ cat /proc/meminfo
>         total:    used:    free:  shared: buffers:  cached:
> Mem:  789250048 781295616  7954432   659456 402890752
> 18446744073478758400
> Swap: 6744576000   282624 6744293376
> MemTotal:       770752 kB
> MemFree:          7768 kB
> MemShared:         644 kB
> Buffers:        393448 kB
> Cached:       4294741680 kB     <------ This is impossible, i think? :-)

Problem appeared in my 2.4.13-ac4 yesterday at home too. It happened to 
me when I was checking health of my HDD - either during 'dd if=/dev/hde1 
of=/dev/null bs=8M', or during copying all files from VFAT (/dev/hde1) 
partition to /dev/null on filesystem level file by file. 

And if we are talking about it, 2.4.13-ac4 here at work reports that
too. But strange thing is that this machine has just 200MB VFAT partition
of no use, and I do not remember that I ever did read from /dev/hd* since
last reboot. Shift-scrolllock does not report any unusual values. 

I'll upgrade to -ac6 and I'll see.
                                                Petr Vandrovec
                                                vandrove@vc.cvut.cz
                                                

Re: Weird /proc/meminfo output on 2.4.13-ac5

[Mike Fedyk]

On Fri, Nov 02, 2001 at 07:34:49PM +0000, Petr Vandrovec wrote:
> On  2 Nov 01 at 14:10, Andreas Franck wrote:
> > $ cat /proc/meminfo
> >         total:    used:    free:  shared: buffers:  cached:
> > Mem:  789250048 781295616  7954432   659456 402890752
> > 18446744073478758400
> > Swap: 6744576000   282624 6744293376
> > MemTotal:       770752 kB
> > MemFree:          7768 kB
> > MemShared:         644 kB
> > Buffers:        393448 kB
> > Cached:       4294741680 kB     <------ This is impossible, i think? :-)
> 
> Problem appeared in my 2.4.13-ac4 yesterday at home too. It happened to 
> me when I was checking health of my HDD - either during 'dd if=/dev/hde1 
> of=/dev/null bs=8M', or during copying all files from VFAT (/dev/hde1) 
> partition to /dev/null on filesystem level file by file. 
> 
> And if we are talking about it, 2.4.13-ac4 here at work reports that
> too. But strange thing is that this machine has just 200MB VFAT partition
> of no use, and I do not remember that I ever did read from /dev/hd* since
> last reboot. Shift-scrolllock does not report any unusual values. 
> 
> I'll upgrade to -ac6 and I'll see.

It won't help.  You'll need a patch that rik has posted a few days ago.

This problem is for 2.4.13, 2.4.13-acX, and 2.4.14pre*.

Latest pre or ac patches don't fix it.

Latest:
pre6
-ac6

Mike

Re: Weird /proc/meminfo output on 2.4.13-ac5

[Hugh Dickins]

On Fri, 2 Nov 2001, Mike Fedyk wrote:
> On Fri, Nov 02, 2001 at 07:34:49PM +0000, Petr Vandrovec wrote:
> > On  2 Nov 01 at 14:10, Andreas Franck wrote:
> > > $ cat /proc/meminfo
> > > Cached:       4294741680 kB     <------ This is impossible, i think? :-)
> > 
> > I'll upgrade to -ac6 and I'll see.
> 
> It won't help.  You'll need a patch that rik has posted a few days ago.
> 
> This problem is for 2.4.13, 2.4.13-acX, and 2.4.14pre*.

I believe that problem only applied to 2.4.13-acX, and arose because
an inappropriate part of 2.4.13 (relating to blockdev in pagecache) crept
into 2.4.13-acX.  2.4.13 and 2.4.14-preX should not need Rik's patch.

Hugh

Re: Weird /proc/meminfo output on 2.4.13-ac5

[Mike Fedyk]

On Fri, Nov 02, 2001 at 10:35:29PM +0000, Hugh Dickins wrote:
> On Fri, 2 Nov 2001, Mike Fedyk wrote:
> > On Fri, Nov 02, 2001 at 07:34:49PM +0000, Petr Vandrovec wrote:
> > > On  2 Nov 01 at 14:10, Andreas Franck wrote:
> > > > $ cat /proc/meminfo
> > > > Cached:       4294741680 kB     <------ This is impossible, i think? :-)
> > > 
> > > I'll upgrade to -ac6 and I'll see.
> > 
> > It won't help.  You'll need a patch that rik has posted a few days ago.
> > 
> > This problem is for 2.4.13, 2.4.13-acX, and 2.4.14pre*.
> 
> I believe that problem only applied to 2.4.13-acX, and arose because
> an inappropriate part of 2.4.13 (relating to blockdev in pagecache) crept
> into 2.4.13-acX.  2.4.13 and 2.4.14-preX should not need Rik's patch.
> 
> Hugh
> 

I am running: $ ud -d
- Uptime for mikef-linux -
Now  : 1 day(s), 03:44:33 running Linux
       2.4.14-pre6+preempt+netdev_random+ext3_0.9.14-2414p5

Check out vmstat:
 0  0  0  49852   3340 180884 4294884956   0   0     1   126  134   249   2
1  97
 1  0  0  49852   3916 180424 4294885152   0   0     0    62  127   254   3
2  95
 0  0  0  49852   3512 180472 4294885496   0   0     0    60  133   267   5
3  91
 0  0  0  49852   4172 179696 4294885660   0   0     0    44  120   228   3
3  94
 0  0  0  49948   4176 178024 4294885640   0  25   120    93  141   362   9
6  85
 0  1  0  49948   3392 174684 4294883668   0  41   249   137  183   415  19
11  70
 0  1  0  49944   3188 176904 4294880964   9  31   293   166  198   450   5
7  89
 0  1  0  50160   4832 177164 4294880100   0   0    76   197  131   395  36
12  52

There you go.  I've also seen a report against 2.4.13 on this list from
Miroslav Zubcic.

Mike

Weird /proc/meminfo output on 2.4.13-ac5

[Andreas Franck]

Hello folks,

since about two days I'm running a machine here with 2.4.13-ac5,
and am quite satisfied with the performance. But now, I noticed
some strange output in /proc/meminfo:

$ cat /proc/meminfo
        total:    used:    free:  shared: buffers:  cached:
Mem:  789250048 781295616  7954432   659456 402890752
18446744073478758400
Swap: 6744576000   282624 6744293376
MemTotal:       770752 kB
MemFree:          7768 kB
MemShared:         644 kB
Buffers:        393448 kB
Cached:       4294741680 kB     <------ This is impossible, i think? :-)
SwapCached:        232 kB
Active:         254160 kB
Inact_dirty:    307272 kB
Inact_clean:        80 kB
Inact_target:   157284 kB
HighTotal:           0 kB
HighFree:            0 kB
LowTotal:       770752 kB
LowFree:          7768 kB
SwapTotal:     6586500 kB
SwapFree:      6586224 kB

The machine is an AMD Athlon 1200, with 768 MB RAM and about 6G swap.
Is there anything more you need to investigate this problem? Then please
don't hesitate to ask.

Greetings,
Andreas