lkml.org 
[lkml]   [1998]   [Apr]   [26]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: enable_ioapic_irq broken in arch/i386/kernel/irq.c
On Sat, 25 Apr 1998, Linus Torvalds wrote:

>Would some io-apic people try out this patch to irq.c (patch against
>2.1.98)? It simplifies the irq handling quite a bit, I'd like to hear
>whether it is stable for you..

The patch seems to work fine here on my Tyan Tomcat IV with 2x Pentium
133's. I'm running it right now and it's been up a half hour.

I did a 'make MAKE="make -j' zImage' on a Linux 2.0.33 kernel tree. Build
progressed fine with a load of 70 (with 10 concurrent 'find /' and 'ping -f
localhost') until it hit the net/ipv4/ directory.

The amount of compiles outweighed my 64MB of RAM and 32 MB of swap and I
reached a swap deadlock. The hard drive wrote quick bursts with the
interval increasing until it did something every 30 seconds. I did a
little checking around during that dead time:

ALT+SysRQ showed:

Free pages: 4048k
Buffer pages: 1964k
Free swap: 0k

Machine was looping around:

c011f848 T shrink_mmap
c011f8a0 <- Here
c011fa3a <- Here
c011fab0 T page_unuse

c01247a8 t swap_out_vma
c0124908 <- Here.
c0124bf4 t swap_out_process
[...]
c01987f0 T __lock_kernel
c0198803 <- Here
c019881f <- Here (These are most likely the second processor.)
c019882c T __delay

A ALT+SysRQ+S synced about 5 seconds later. I brought the machine back to
life with a ALT+SysRQ+K on the 'make -j' terminal and all has been well
since.

So summary:

* Patch seems to work ok.

* 'make -j zImage' will deadlock when you run out of swap, even if you have
free memory, instead of the normal 'Virtual memory exhausted.'

-George


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu

\
 
 \ /
  Last update: 2005-03-22 13:42    [W:0.107 / U:0.128 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site