HI !

Kernel      : 2.3.34 
HW-Platform : DP-264 (Tsunami)
RAM         : 1GB
SWAP        : 512MB (4 x 128MB)
OS-Version  : SuSE 6.3-AXP
Compileropts: SMP, DP-264, BSD-Disklabels (also tryed using Generic-Alpha)

 I configured 4 swap partitions of 128MB each it shows 512MB RAM 
 if I look at /proc/meminfo all seems ok , 

        total:    used:    free:  shared: buffers:  cached:
Mem:  1044414464 971636736 72777728        0   507904 33415168
Swap: 568295424   262144 568033280
MemTotal:   1019936 kB
MemFree:      71072 kB
MemShared:        0 kB
Buffers:        496 kB
Cached:       32632 kB
HighTotal:        0 kB
HighFree:         0 kB
LowTotal:   1019936 kB
LowFree:      71072 kB
SwapTotal:   554976 kB
SwapFree:    554720 kB


And top also seams ok at the top lines , but the memory percentage of used 
memory indicates that it is assuming a total ram of 2GB ! and as soon as
*ANY* process uses more than the first 128MB of swap space , the box freezes
without any further comment.

4:14pm  up  5:30,  7 users,  load average: 1.00, 1.00, 1.00
58 processes: 56 sleeping, 2 running, 0 zombie, 0 stopped
CPU states: 64.7% user, 37.7% system, 98.6% nice,  0.0% idle
Mem:  1019936K av, 952352K used,  67584K free,      0K shrd,    576K buff
Swap: 554976K av,    256K used, 554720K free                 34776K cached
  PID USER     PRI  NI  SIZE  RSS SHARE STAT  LIB %CPU %MEM   TIME COMMAND
  431 dhobbs    19   5  868M 868M  3408 R N  866M 98.6 43.5 249:36 vasp
  586 root       2   0  1624 1624  1288 R    1520  3.9  0.0   0:00 top


 the problem I guess is the BSD-disklable support.

Command (m for help): 
BSD disklabel command (m for help): 
8 partitions:
#        size   offset    fstype   [fsize bsize   cpg]
  a:   260096     4096      ext2                        # (Cyl.    3 - 129)
  b:  1576960   264192      ext2                        # (Cyl.  130 - 899)
  c:  4098048  1841152      ext2                        # (Cyl.  900 - 2900)
  d:   276480  5939200      swap                        # (Cyl. 2901 - 3035)
  e:   278528  6215680      swap                        # (Cyl. 3036 - 3171)
  f:   276480  6494208      swap                        # (Cyl. 3172 - 3306)
  g:   278528  6770688      swap                        # (Cyl. 3307 - 3442)
  h: 28794880  7049216      ext2                        # (Cyl. 3443 - 17502)

BSD disklabel command (m for help): 
 
output from dmesg:
------------------
....
Starting kswapd v1.6
....
Partition check:
 sda: sda1 sda2 sda3 sda4 sda5 sda6 sda7 sda8
VFS: Mounted root (ext2 filesystem) readonly.
Freeing unused kernel memory: 208k freed
Adding Swap: 138232k swap-space (priority -1)
Adding Swap: 139256k swap-space (priority -2)
Adding Swap: 138232k swap-space (priority -3)
Adding Swap: 139256k swap-space (priority -4)

 So my assumtion is that some part of the kernel is calculating the swapspace
 with  278528*512k wich gives me the actual 139256 but other parts seam to 
 calculate 278528*1024k (..top %MEM ) and when this first swap partitions
 physical end is reached at 128MB it messes up , accessing a swap area that 
 does not exist...

any hint where to start diging ??
are any problems with BSD-Disklabels as such known ??
or are my assumptions completly wrong ?

thx
hofrat
*------------------------------------------------------------------------*
*THERE           {__}                         Universitaet Wien          *
*IS              oo )                      Inst. f. Materialphysik       *
*LIFE           O_   `_-------.              Mc  Guire  Nicholas         *
*IN               (-.         ,-\                                        *
*THE               || )---<  )            [EMAIL PROTECTED] *
*NET !             |||     | |          privat: [EMAIL PROTECTED]         *
*                                     [EMAIL PROTECTED]   *
*------------------------------------------------------------------------*
*ASCII-COW (C) 1996-2001 [EMAIL PROTECTED]                                *
*------------------------------------------------------------------------*

Reply via email to