[uml-devel] Fwd: [uml-user] Host 2.6.2/3 + CFQ + BlaisorBlade patches == host panic

[BlaisorBlade <blaisorblade_spam@yahoo.it>]

Hi Ingo, could you please take a look at this report? With CONFIG_SMP enabled 
and the skas patch applied (last version, which includes all your changes) 
the kernel is BUGging during switch_mm.

I've looked at include/asm-i386/mmu_context.h and I saw this code:

#ifdef CONFIG_SMP
        else {
                cpu_tlbstate[cpu].state = TLBSTATE_OK;
                BUG_ON(cpu_tlbstate[cpu].active_mm != next); /*!!!!This is the 
bug you hit*/
...

The kernel is complaining because prev == next but prev != 
cpu_tlbstate[cpu].active_mm. switch_mm is called in schedule.c (looking at 
the EIP value below, which is inside schedule()) and its params are actually 
prev->active_mm and next->mm. So it BUGged because prev->active_mm was false, 
or because prev was not running when it was scheduled away (or was running 
onto another CPU). However I cannot understand anything more

Maybe prev->active_mm had just been changed by ptrace but 
cpu_tlbstate[cpu].active_mm had not been updated? I.e. the "child" task in 
ptrace had not been re-scheduled in - was it maybe running somewhere else?

----------  Forwarded message  ----------

Subject: [uml-user] Host 2.6.2/3 + CFQ + BlaisorBlade patches == host panic
Date: 22:04, venerdì 20 febbraio 2004
From: "Christopher S. Aker" <caker@theshore.net>
To: <user-mode-linux-user@lists.sourceforge.net>
Cc: "BlaisorBlade" <blaisorblade_spam@yahoo.it>

Hello,

I've been testing 2.6 and SKAS and receive the following BUG on both 2.6.2
 and 2.6.3.  Patches applied are (in order):

*
http://www.user-mode-linux.org/~blaisorblade/patches/2.6.2-v1/uml-patch-2.6.2
-0.bz2 *
 http://www.user-mode-linux.org/~blaisorblade/patches/2.6.2-v1/00-Combo-v1.bz
2 *
http://www.user-mode-linux.org/~blaisorblade/patches/2.6.2-v1-host-skas2/00-2
.6.2-v1-host-SKAS2.patch.bz2 * the CFQ patch from the -mm tree
* patch-2.6.3 (only on the 2.6.3 kernel, obviously)

kernel BUG at include/asm/mmu_context.h:53!
invalid operand: 0000 [#1]
CPU:    2
EIP:    0060:[<c011dff1>]    Not tainted
EFLAGS: 00010083
EIP is at schedule+0x53e/0x6c5
eax: 00000100   ebx: f57e0d00   ecx: 00000002   edx: f57e0d00
esi: f640c6b0   edi: c3da1c80   ebp: f7f97f78   esp: f7f97f24
ds: 007b   es: 007b   ss: 0068
Process swapper (pid: 0, threadinfo=f7f96000 task=f7f9ece0)
Stack: f640c6b0 3ee9c9d5 00000185 f7f97f30 e3cd6000 00000200 00000001
 c0451b88 f640c6b0 3ee9c9d5 00000185 f640c6b0 c3da1c80 0001ab57 3ee9cded
 00000185 f7f9ece0 f7f9eeac f7f96000 f7f96000 c010882e f7f96000 c010ac89
 f7f96000 Call Trace:
 [<c010882e>] default_idle+0x0/0x2d
 [<c010ac89>] need_resched+0x27/0x32
 [<c010882e>] default_idle+0x0/0x2d
 [<c0108858>] default_idle+0x2a/0x2d
 [<c01088cc>] cpu_idle+0x37/0x40
 [<c012314e>] printk+0x188/0x1d0
 [<c045c51d>] print_cpu_info+0x86/0xd2

Code: 0f 0b 35 00 34 3a 3a c0 eb c5 89 5e 6c f0 ff 43 18 b8 00 e0

Does this look like it could be caused by the SKAS patches, or would you
 suspect the CFQ patch?  The box seems to run fine, but once I fire up a few
 UMLs and start banging around, the host will BUG.

Thanks,
-Chris

-- 
Paolo Giarrusso, aka Blaisorblade
Linux registered user n. 292729



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id\x1356&alloc_id438&opÌk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel