Re: Large ramdisk crashes system

Re: Large ramdisk crashes system

[Marcelo Tosatti]

On Thu, 7 Jun 2001, Paul Buder wrote:

> I am trying to create a system which boots off of a cd and has no hard
> disks.  So it needs ramdisks.  But I haven't had much luck creating
> large ones.
> 
> I tried on two different boxes.  In both cases the kernel is 2.4.5 with
> 'Simple RAM-based file system support' turned on.
> 
> One box is a dual Pentium 750 with a gig of ram in it.  I had the
> kernel 'Default RAM disk size' set to 800000 for this box.  I issued
> the following commands.
> 
> mkfs /dev/ram0 400000
> mount /dev/ram0 /mnt
> dd if=/dev/zero of=/mnt/junk bs=1024 count=500000
> 
> This is fine, dd creates a 400 meg file, reports there isn't enough
> space and exits.  But if I change the first line to
> 
> mkfs /dev/ram0 500000
> 
> I'm essentially crashed.  I can ping the box and switch between virtual
> terminals but that's it.  Any program that was running on the other
> virtual terminals is frozen (as in top, tail, login).  The dd is frozen
> and can't be control-c'd.  so I can't do anything other than powercycle.
> I should have at least 400 megs of ram left for the system so I don't
> get it.
> 
> I tried the same thing on a 128 meg box.  The results were similar.  A 40
> meg ram disk worked.  A 60 meg ram disk crashed the box.  The numbers
> seem a little odd since in both cases the magic threshold seems to be
> roughly 40% of ram.
> 
> I get no messages in the system logfiles nor an oops on the screen.
> 
> Any ideas?

Can you get the (traced by ksymoops) backtrace of dd and kswapd
everything is locked? 

You can do that with sysrq. 

Large ramdisk crashes system

[Paul Buder]

I am trying to create a system which boots off of a cd and has no hard
disks.  So it needs ramdisks.  But I haven't had much luck creating
large ones.

I tried on two different boxes.  In both cases the kernel is 2.4.5 with
'Simple RAM-based file system support' turned on.

One box is a dual Pentium 750 with a gig of ram in it.  I had the
kernel 'Default RAM disk size' set to 800000 for this box.  I issued
the following commands.

mkfs /dev/ram0 400000
mount /dev/ram0 /mnt
dd if=/dev/zero of=/mnt/junk bs=1024 count=500000

This is fine, dd creates a 400 meg file, reports there isn't enough
space and exits.  But if I change the first line to

mkfs /dev/ram0 500000

I'm essentially crashed.  I can ping the box and switch between virtual
terminals but that's it.  Any program that was running on the other
virtual terminals is frozen (as in top, tail, login).  The dd is frozen
and can't be control-c'd.  so I can't do anything other than powercycle.
I should have at least 400 megs of ram left for the system so I don't
get it.

I tried the same thing on a 128 meg box.  The results were similar.  A 40
meg ram disk worked.  A 60 meg ram disk crashed the box.  The numbers
seem a little odd since in both cases the magic threshold seems to be
roughly 40% of ram.

I get no messages in the system logfiles nor an oops on the screen.

Any ideas?

Re: Large ramdisk crashes system

[David Woodhouse]

paulb@aracnet.com said:
> the kernel is 2.4.5 with 'Simple RAM-based file system support' turned on.

> I issued the following commands.

> mkfs /dev/ram0 400000
> mount /dev/ram0 /mnt
> dd if=/dev/zero of=/mnt/junk bs=1024 count=500000 

Why turn on ramfs if you're not going to use it? 

 mount -t ramfs none /mnt/junk

Use the one in the -ac tree and you get resource limits, which will be 
useful. The VM will still be broken, but you should get away with a little 
more.

--
dwmw2

Re: Large ramdisk crashes system

[Paul Buder]

On Thu, 7 Jun 2001, Marcelo Tosatti wrote:

> On Thu, 7 Jun 2001, Paul Buder wrote:
>
> > I am trying to create a system which boots off of a cd and has no hard
> > disks.  So it needs ramdisks.  But I haven't had much luck creating
> > large ones.

[ explanation of large ram disks crashing the system edited out
for brevity]
>
> Can you get the (traced by ksymoops) backtrace of dd and kswapd
> everything is locked?
>
> You can do that with sysrq.


I copied the sysreq-t screen to paper and then typed it up and fed
it to ksymoops.  I get some errors since this kernel has module support
turned off. I also get a message from ksymoops saying
Warning (Oops_read): Code line not seen, dumping what data is available
which I'm not sure of the meaning. So anyway, my oops file and
the output from ksymoops follow.  Hopefully I've done this right.
If there is anything else I can provide let me know.


kswapd    R f7d5abfc   0     3      1        (L-TLB)       4     2
Call Trace: c010fd6b>] [<c010fd6b>] [<c0162bab>] [<c01964c1>] [<c019d90e>] [<c019d804>] [<c0117d8c>]
[<c010e526>] [<c01247e4>] [<c01247e4>] [<c010e030>] [<c01247e4>] [<c010deab>] [<c01260aa>] [<c0126115>]
[<c0126299>] [<c0126349>] [<c01263dc>] [<c011153f>] [<c012727d>] [<c0127459>] [<c01274d2>] [<c012753b>]
[<c01056d0>]
dd        R current     756   249   187        (NOTLB)
Call Trace: [<c015c184>] [<c01964c1>] [<c019ddab>] [<c019d804>] [<c0194799>] [<c01974f2>] [<c011043e>]
[<c012e4e5>] [<c0173b8b>] [<c0173bbf>] [<c0173bd9>] [<c01964c1>] [<c019ddab>] [<c019d804>] [<c0197499>]
[<c01974f2>] [<c020e742>] [<c01e81b0>] [<c020956b>] [<c01df693>] [<c01d923d>] [<c01e8253>] [<c01df143>]
[<c01e6da2>] [<c01e81b0>] [<c01e81a9>] [<c01df143>] [<c01e7a5f>] [<c01e819c>] [<c0201c1c>] [<c0201a2c>]
[<c0201c30>] [<c0202322>] [<c017184c>] [<c016be8f>] [<c016b159>] [<c01150ea>] [<c012f88a>] [<c017184c>]
[<c01c5df0>] [<c0167ff2>] [<c016b464>] [<c0112d86>] [<c0112d86>] [<f8800000>] [<c01071d8>] [<c01071d8>]
[<c0107207>] [<c0111033>] [<c01110ee>] [<c017307f>] [<c0171747>] [<c01728bd>] [<c0172948>] [<c01085f0>]
[<c01087d7>] [<c0106f40>] [<c021ce04>] [<c0122880>] [<c012d7aa>] [<c0106e67>]



ksymoops 2.4.1 on i686 2.4.5.  Options used
     -V (default)
     -k /proc/ksyms (default)
     -l /proc/modules (default)
     -o /lib/modules/2.4.5/ (default)
     -m /usr/src/linux/System.map (specified)

Error (regular_file): read_ksyms stat /proc/ksyms failed
No modules in ksyms, skipping objects
No ksyms, skipping lsmod
Call Trace: c010fd6b>] [<c010fd6b>] [<c0162bab>] [<c01964c1>] [<c019d90e>] [<c019d804>] [<c0117d8c>]
[<c010e526>] [<c01247e4>] [<c01247e4>] [<c010e030>] [<c01247e4>] [<c010deab>] [<c01260aa>] [<c0126115>]
[<c0126299>] [<c0126349>] [<c01263dc>] [<c011153f>] [<c012727d>] [<c0127459>] [<c01274d2>] [<c012753b>]
[<c01056d0>]
Call Trace: [<c015c184>] [<c01964c1>] [<c019ddab>] [<c019d804>] [<c0194799>] [<c01974f2>] [<c011043e>]
[<c012e4e5>] [<c0173b8b>] [<c0173bbf>] [<c0173bd9>] [<c01964c1>] [<c019ddab>] [<c019d804>] [<c0197499>]
[<c01974f2>] [<c020e742>] [<c01e81b0>] [<c020956b>] [<c01df693>] [<c01d923d>] [<c01e8253>] [<c01df143>]
[<c01e6da2>] [<c01e81b0>] [<c01e81a9>] [<c01df143>] [<c01e7a5f>] [<c01e819c>] [<c0201c1c>] [<c0201a2c>]
[<c0201c30>] [<c0202322>] [<c017184c>] [<c016be8f>] [<c016b159>] [<c01150ea>] [<c012f88a>] [<c017184c>]
[<c01c5df0>] [<c0167ff2>] [<c016b464>] [<c0112d86>] [<c0112d86>] [<f8800000>] [<c01071d8>] [<c01071d8>]
[<c0107207>] [<c0111033>] [<c01110ee>] [<c017307f>] [<c0171747>] [<c01728bd>] [<c0172948>] [<c01085f0>]
[<c01087d7>] [<c0106f40>] [<c021ce04>] [<c0122880>] [<c012d7aa>] [<c0106e67>]
Warning (Oops_read): Code line not seen, dumping what data is available

Trace; c010fd6b <reschedule_idle+63/214>
Trace; c010e526 <smp_apic_timer_interrupt+ea/fc>
Trace; c01247e4 <do_ccupdate_local+0/34>
Trace; c01247e4 <do_ccupdate_local+0/34>
Trace; c010e030 <smp_call_function+8c/c4>
Trace; c01247e4 <do_ccupdate_local+0/34>
Trace; c010deab <flush_tlb_page+7b/84>
Trace; c01260aa <try_to_swap_out+9a/1b4>
Trace; c0126115 <try_to_swap_out+105/1b4>
Trace; c0126299 <swap_out_pmd+d5/f0>
Trace; c0126349 <swap_out_vma+95/d8>
Trace; c01263dc <swap_out_mm+50/7c>
Trace; c011153f <mmput+13/50>
Trace; c012727d <refill_inactive_scan+f1/160>
Trace; c0127459 <refill_inactive+6d/a4>
Trace; c01274d2 <do_try_to_free_pages+42/58>
Trace; c012753b <kswapd+53/e0>
Trace; c01056d0 <kernel_thread+28/38>
Trace; c015c184 <do_tty_hangup+0/314>
Trace; c01964c1 <ide_set_handler+61/74>
Trace; c019ddab <do_rw_disk+12f/2e8>
Trace; c019d804 <read_intr+0/11c>
Trace; c0194799 <drm_dma_takedown+1d9/1e4>
Trace; c01974f2 <start_request+182/1f0>
Trace; c011043e <schedule+3aa/5ac>
Trace; c012e4e5 <__wait_on_buffer+85/94>
Trace; c0173b8b <__make_request+15b/6bc>
Trace; c0173bbf <__make_request+18f/6bc>
Trace; c0173bd9 <__make_request+1a9/6bc>
Trace; c01964c1 <ide_set_handler+61/74>
Trace; c019ddab <do_rw_disk+12f/2e8>
Trace; c019d804 <read_intr+0/11c>
Trace; c0197499 <start_request+129/1f0>
Trace; c01974f2 <start_request+182/1f0>
Trace; c020e742 <ip_nat_fn+1a2/1ac>
Trace; c01e81b0 <ip_finish_output2+0/dc>
Trace; c020956b <ip_refrag+23/54>
Trace; c01df693 <qdisc_restart+13/178>
Trace; c01d923d <dev_queue_xmit+10d/258>
Trace; c01e8253 <ip_finish_output2+a3/dc>
Trace; c01df143 <nf_hook_slow+11b/180>
Trace; c01e6da2 <ip_output+126/130>
Trace; c01e81b0 <ip_finish_output2+0/dc>
Trace; c01e81a9 <output_maybe_reroute+d/14>
Trace; c01df143 <nf_hook_slow+11b/180>
Trace; c01e7a5f <ip_build_xmit+2df/36c>
Trace; c01e819c <output_maybe_reroute+0/14>
Trace; c0201c1c <icmp_reply+16c/188>
Trace; c0201a2c <icmp_glue_bits+0/84>
Trace; c0201c30 <icmp_reply+180/188>
Trace; c0202322 <icmp_echo+4a/54>
Trace; c017184c <handle_scancode+2ac/2e8>
Trace; c016be8f <poke_blanked_console+57/5c>
Trace; c016b159 <console_softint+81/d4>
Trace; c01150ea <tasklet_action+4e/7c>
Trace; c012f88a <bread+16/60>
Trace; c017184c <handle_scancode+2ac/2e8>
Trace; c01c5df0 <vgacon_cursor+1a8/1b0>
Trace; c0167ff2 <set_cursor+6e/80>
Trace; c016b464 <vt_console_print+2b8/2d0>
Trace; c0112d86 <printk+176/184>
Trace; c0112d86 <printk+176/184>
Trace; f8800000 <END_OF_CODE+384f734c/????>
Trace; c01071d8 <show_trace+88/a8>
Trace; c01071d8 <show_trace+88/a8>
Trace; c0107207 <show_trace_task+f/14>
Trace; c0111033 <show_task+15b/160>
Trace; c01110ee <show_state+3e/58>
Trace; c017307f <handle_sysrq+15f/230>
Trace; c0171747 <handle_scancode+1a7/2e8>
Trace; c01728bd <handle_kbd_event+121/190>
Trace; c0172948 <keyboard_interrupt+1c/28>
Trace; c01085f0 <handle_IRQ_event+58/84>
Trace; c01087d7 <do_IRQ+a7/f8>
Trace; c0106f40 <ret_from_intr+0/20>
Trace; c021ce04 <stext_lock+1514/6de8>
Trace; c0122880 <generic_file_write+498/540>
Trace; c012d7aa <sys_write+8e/c4>
Trace; c0106e67 <system_call+37/40>


1 warning and 1 error issued.  Results may not be reliable.

Re: Large ramdisk crashes system

[Paul Buder]

On Fri, 8 Jun 2001, David Woodhouse wrote:

>
> paulb@aracnet.com said:
> > the kernel is 2.4.5 with 'Simple RAM-based file system support' turned on.
>
> > I issued the following commands.
>
> > mkfs /dev/ram0 400000
> > mount /dev/ram0 /mnt
> > dd if=/dev/zero of=/mnt/junk bs=1024 count=500000
>
> Why turn on ramfs if you're not going to use it?
>
Actually I experimented with both ext2 and ramfs, getting similar
results.  I forgot to mention that in the post though.