[uml-devel] Segfault on UML load

[uml-devel] Segfault on UML load

[Charles A. Clinton]

When I build a UML kernel on my Fedora Core 1 machine, it segfaults when 
it tries to load. It looks to me like changes in binutils to support 
exec-shield have added a STACK program header at 0x00000000, causing the 
image to appear to be 2.5G in size.

Is this a:
    1) known and solved
    2) an open issue that I need to work on a fix for
    3) nobody cares
kind of problem?

(My guess is that adding some commands to arch/um/ld.link.in to disappear 
or move the STACK program header is the right way to fix it)

-- Charles


-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] Segfault on UML load

[BlaisorBlade]

Alle 01:56, giovedì 1 aprile 2004, Charles A. Clinton ha scritto:
> When I build a UML kernel on my Fedora Core 1 machine, it segfaults when
> it tries to load.
There are reports of this, but was related either to NPTL or to exec-shield 
(and people succeeded in solving it).

> It looks to me like changes in binutils to support
> exec-shield have added a STACK program header at 0x00000000, causing the
> image to appear to be 2.5G in size.
Nobody has ever had this clue... I've no idea of those changes.

> Is this a:
>     1) known and solved
>     2) an open issue that I need to work on a fix for
>     3) nobody cares
> kind of problem?
>
> (My guess is that adding some commands to arch/um/ld.link.in to disappear
> or move the STACK program header is the right way to fix it)

Well, see this for the solution everybody used:

http://toast.debian.net/~may/umlproject/bugs.html

If you succeed in doing the patch you suggest, then post the patch. If you 
think you are right but aren't able to do the fix, wait for answers from 
somebody smarter than me(especially M A Young who hacks UML on Fedora) or 
post evidence/explanations about your problem.

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



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] Segfault on UML load

[Charles A. Clinton]

On Wed, 31 Mar 2004, I wrote:

> When I build a UML kernel on my Fedora Core 1 machine, it segfaults when 
> it tries to load.

Blaisorblade pointed me to a 2.6 patch set which provides the answer.
The link.ld script puts an empty 4K page at the front of the UML image.
The Fedora glibc expects the elf program headers to be at the start of the
image and searched them form information regarding stack executability. 
The empty page is not mapped, and the search segfaults. I suspect that 
there is a bug in binfmt_elf that is not taking into account the load 
offset when setting up the aux vector, but the code is so convoluted there 
that I am mostly dazed.

Anyway, the following patch (adapted from
http://toast.debian.net/~may/umlproject/unpackaged/align.patch) when 
applied on top of uml-patch-2.4.22-1, fixes the problem. It should work
for any of the 2.4.x patches.

Thanks to Blaisorblade for pointing me to the UML for Fedora page.

diff -Naur linux-2.4.22/arch/um/dyn_link.ld.in linux-2.4.22.works/arch/um/dyn_link.ld.in
--- linux-2.4.22/arch/um/dyn_link.ld.in	2004-04-01 11:26:04.000000000 -0800
+++ linux-2.4.22.works/arch/um/dyn_link.ld.in	2004-04-01 11:32:35.000000000 -0800
@@ -8,7 +8,6 @@
 {
   . = START() + SIZEOF_HEADERS;
   .interp         : { *(.interp) }
-  . = ALIGN(4096);
   __binary_start = .;
   . = ALIGN(4096);		/* Init code and data */
   _stext = .;
diff -Naur linux-2.4.22/arch/um/link.ld.in linux-2.4.22.works/arch/um/link.ld.in
--- linux-2.4.22/arch/um/link.ld.in	2004-04-01 11:26:07.000000000 -0800
+++ linux-2.4.22.works/arch/um/link.ld.in	2004-04-01 11:32:19.000000000 -0800
@@ -6,7 +6,6 @@
 {
   . = START() + SIZEOF_HEADERS;
 
-  . = ALIGN(4096);
   __binary_start = .;
 ifdef(`MODE_TT', `
   .thread_private : {


-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel

Re: [uml-devel] Segfault on UML load

[BlaisorBlade]

Alle 00:16, venerdì 2 aprile 2004, Charles A. Clinton ha scritto:
> Anyway, the following patch (adapted from
> http://toast.debian.net/~may/umlproject/unpackaged/align.patch) when
> applied on top of uml-patch-2.4.22-1, fixes the problem. It should work
> for any of the 2.4.x patches.
Ahh, ok. I did forgot that one... anyway, always state patch release! That 
patch is very old; newer patches have this bug fixed (at least the 2.4.24 one 
).

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



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id\x1470&alloc_id638&opÌk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel