* Re: 2.6.0 kernel panic
2003-12-30 6:42 ` Linus Torvalds
@ 2003-12-30 7:07 ` Murray J. Root
2003-12-31 3:53 ` Isaac Claymore
1 sibling, 0 replies; 6+ messages in thread
From: Murray J. Root @ 2003-12-30 7:07 UTC (permalink / raw)
To: linux-kernel
On Mon, Dec 29, 2003 at 10:42:06PM -0800, Linus Torvalds wrote:
>
>
> On Mon, 29 Dec 2003, Murray J. Root wrote:
> > On Sat, Dec 27, 2003 at 09:07:59PM -0500, Murray J. Root wrote:
> > > P4 2GHz
> > > ASUS P4S533 mainboard
> > > 1G PC2700 RAM
> > > GF2 GTS video using nv driver
> > > 2.6.0 compiled with gcc 3.3.2
> > >
> > > At boot kernel gets:
> > > INIT: cannot execute "/etc/rc.d/rc.sysinit"
> > > then panic.
> > >
> > > Same configuration for 2.6.0-test11 and earlier works fine.
> > >
> >
> > To answer myself, I did a diff between 2.6.0-test11 and 2.6.0. Found this:
>
> Sounds like one of the partitions that has the executable script loader is
> mounted with "noexec".
>
> On most systems, /etc/rc.d/rc.sysinit is a bash script, and explicitly
> points to /bin/bash. Check "ldd /bin/bash", and verify that all the
> libraries (and /bin itself, of course) are mounted on executable
> filesystems.
>
> That would be a bug that 2.6.0 uncovers.
>
Thought of that. What does the initial mounting of / ? If it's using /etc/fstab
then that isn't the problem - I have no partitions mounting noexec.
/bin and /lib are on the / partition. If it's a script then it's from Mandrake
and I'll have to hunt it down and send them a patch.
Not using devfs or udev. No initrd. All components built-in although module
support is enabled.
/etc/fstab is included below.
--
Murray J. Root
[root@Master source]# cat /etc/fstab
/dev/hda1 /mnt/windows vfat iocharset=iso8859-1,codepage=850,umask=0 0 0
/dev/hda5 / reiserfs notail 1 1
/dev/hda6 /usr/local reiserfs notail 1 2
/dev/hda7 /mirror reiserfs notail 1 2
/dev/hda8 /home reiserfs notail 1 2
/dev/hda9 swap swap defaults 0 0
/dev/hda10 /common reiserfs notail 1 2
/dev/hdc /mnt/cdrom iso9660 noauto,nosuid,ro,umask=0,user,nodev 0 0
/dev/hdd /mnt/cdrom2 iso9660 codepage=850,iocharset=iso8859-1,noauto,nosuid,ro,umask=0,user,nodev 0 0
/dev/sda1 /mnt/disk vfat noauto,rw,user,umask=0 0 0
none /dev/pts devpts mode=0620 0 0
none /proc proc defaults 0 0
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: 2.6.0 kernel panic
2003-12-30 6:42 ` Linus Torvalds
2003-12-30 7:07 ` Murray J. Root
@ 2003-12-31 3:53 ` Isaac Claymore
1 sibling, 0 replies; 6+ messages in thread
From: Isaac Claymore @ 2003-12-31 3:53 UTC (permalink / raw)
To: linux-kernel; +Cc: Marcelo Tosatti
[-- Attachment #1: Type: text/plain, Size: 1525 bytes --]
On Mon, Dec 29, 2003 at 10:42:06PM -0800, Linus Torvalds wrote:
>
>
> On Mon, 29 Dec 2003, Murray J. Root wrote:
> > On Sat, Dec 27, 2003 at 09:07:59PM -0500, Murray J. Root wrote:
> > > P4 2GHz
> > > ASUS P4S533 mainboard
> > > 1G PC2700 RAM
> > > GF2 GTS video using nv driver
> > > 2.6.0 compiled with gcc 3.3.2
> > >
> > > At boot kernel gets:
> > > INIT: cannot execute "/etc/rc.d/rc.sysinit"
> > > then panic.
> > >
> > > Same configuration for 2.6.0-test11 and earlier works fine.
> > >
> >
> > To answer myself, I did a diff between 2.6.0-test11 and 2.6.0. Found this:
>
> Sounds like one of the partitions that has the executable script loader is
> mounted with "noexec".
>
> On most systems, /etc/rc.d/rc.sysinit is a bash script, and explicitly
> points to /bin/bash. Check "ldd /bin/bash", and verify that all the
> libraries (and /bin itself, of course) are mounted on executable
> filesystems.
>
> That would be a bug that 2.6.0 uncovers.
I've noticed that 2.4.23 lacks this additional check, and included a
patch against it.
Marcelo, does this trivial stuff look good enough to be applied?
-Isaac
>
> Linus
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
--
Regards, Isaac
() ascii ribbon campaign - against html e-mail
/\ - against microsoft attachments
[-- Attachment #2: mmap.do_mmap_pgoff.diff --]
[-- Type: text/plain, Size: 670 bytes --]
--- mmap.c 2003-12-05 09:39:09.000000000 +0800
+++ mmap.c.execproto 2003-12-31 11:25:12.000000000 +0800
@@ -17,6 +17,7 @@
#include <asm/uaccess.h>
#include <asm/pgalloc.h>
+#include <linux/mount.h>
/*
* WARNING: the debugging will use recursive algorithms so never enable this
@@ -400,8 +401,13 @@ unsigned long do_mmap_pgoff(struct file
int error;
rb_node_t ** rb_link, * rb_parent;
- if (file && (!file->f_op || !file->f_op->mmap))
- return -ENODEV;
+ if (file) {
+ if (!file->f_op || !file->f_op->mmap)
+ return -ENODEV;
+
+ if ((prot & PROT_EXEC) && (file->f_vfsmnt->mnt_flags & MNT_NOEXEC))
+ return -EPERM;
+ }
if (!len)
return addr;
^ permalink raw reply [flat|nested] 6+ messages in thread