Re: [uml-devel] Re: 2.6.9-bb3 creates zombie threads in SKAS

[Bodo Stroesser <bstroesser@fujitsu-siemens.com>]

[-- Attachment #1: Type: text/plain, Size: 632 bytes --]

Christopher S. Aker wrote:
> I've been able to narrow this down to the "signal-delivery" patch.  2.6.10-rc2-mm3
> without the patch is fine, whereas adding the signal-delivery patch and you'll get
> mysql zombies inside the UML.  This happens on a variety of hosts, including
> 2.6.10-rc2-bk7 + skas-v7.
> 
> I just noticed 2.6.10-rc2-mm4 is out, and the patch there is named
> "uml-redo-the-signal-delivery-mechanism.patch".
> 
> -Chris
> 
Have found a bug in the signal-delivery patch a few days ago.
Could you please try the appended patch? (It's relevant for usage of RTSIGs
only, don't know, whether this could help you)

Bodo

[-- Attachment #2: extramask.patch --]
[-- Type: text/x-diff, Size: 1287 bytes --]


From: Bodo Stroesser <bstroesser@fujitsu-siemens.com>

Restoring of current->blocked in sys_sigreturn is wrong.
The first (long ) of the mask correctly is fetched from sc->oldmask.
The further longs again come from there, but correctly should be
taken from frame->extramask.

Signed-off-by: Bodo Stroesser <bstroesser@fujitsu-siemens.com>
---


diff -puN arch/um/sys-i386/signal.c~extramask arch/um/sys-i386/signal.c
--- linux-2.6.10-rc2/arch/um/sys-i386/signal.c~extramask	2004-11-26 14:03:10.770096974 +0100
+++ linux-2.6.10-rc2-root/arch/um/sys-i386/signal.c	2004-11-26 14:11:03.264403385 +0100
@@ -307,11 +307,12 @@ long sys_sigreturn(struct pt_regs regs)
 	struct sigframe __user *frame = (struct sigframe *)(sp - 8);
 	sigset_t set;
 	struct sigcontext __user *sc = &frame->sc;
-	unsigned long __user *mask = &sc->oldmask;
+	unsigned long __user *oldmask = &sc->oldmask;
+	unsigned long __user *extramask = &frame->extramask;
 	int sig_size = (_NSIG_WORDS - 1) * sizeof(unsigned long);
 
-	if(copy_from_user(&set.sig[0], mask, sizeof(&set.sig[0])) ||
-	   copy_from_user(&set.sig[1], mask, sig_size))
+	if(copy_from_user(&set.sig[0], oldmask, sizeof(&set.sig[0])) ||
+	   copy_from_user(&set.sig[1], extramask, sig_size))
 		goto segfault;
 
 	sigdelsetmask(&set, ~_BLOCKABLE);
_