Bug 231290

Won't be able to resolve this today, so just use yesterday's tree until I can
fix it.

The backtrace:

dcantrel@mortise:~/anaconda-debuginfo-11.2.0.31-1/usr/lib/debug/usr/lib/anaconda
-runtime/loader$ eu-addr2line -e loader.debug 
0x4000000000000530
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1255
0xa0000000000107e0
??:0
0x400000000032c600
/usr/src/debug/anaconda-11.2.0.31/stubs/unicode-lite.c:29
0x400000000003a180
/usr/src/debug/anaconda-11.2.0.31/loader2/net.c:435
0x4000000000002c20
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1102
0x4000000000004f90
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1557
0x40000000002b8af0
/usr/src/debug/anaconda-11.2.0.31/stubs/unicode-lite.c:29
0x4000000000000280
??:0

None of those lines were touched in yesterday's build, so I need to do some more
digging.

same on ppc with the ppc64.img :

loader received SIGSEGV!  Backtrace:                                           
[0x1000037c]
[0x100344]
[0x8000]
[0x80808080]
[0x100185fc]
[0x1001ee5c]
[0x10017bf8]
[0x10018334]
[0x100018ec]
[0x1000292c]
install exited abnormally [1/1] 

with "anaconda installer init version 11.2.0.31 starting"

Backtrace for ppc64 boot (for my own reference later):

dcantrel@mortise:~/anaconda-debuginfo-11.2.0.31-1/usr/lib/debug/usr/lib/anaconda
-runtime/loader$ eu-addr2line -e loader.debug 
0x1000037c
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:172
0x100344
??:0
0x8000
??:0
0x80808080
??:0
0x100185fc
/usr/src/debug/anaconda-11.2.0.31/loader2/nfsinstall.c:361
0x1001ee5c
/usr/src/debug/anaconda-11.2.0.31/loader2/net.c:398
0x10017bf8
/usr/src/debug/anaconda-11.2.0.31/loader2/hdinstall.c:257
0x10018334
/usr/src/debug/anaconda-11.2.0.31/loader2/hdinstall.c:226
0x100018ec
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:962
0x1000292c
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1425

OK, so the good news is I don't think it's related to just 64-bit platforms.  It
seems that loaderData->ip is NULL under certain circumstances.  Working on
reproducing that locally, but I think the fix will be pretty simple to some
changes I made a couple days ago.

Created attachment 149596 [details]
Patch to loader to handle loaderData->ip being NULL

This is what I think is needed to fix the problem.  I am building new
initrd.img files for ia64 and ppc64 and will attach those to this bug report so
you can test.  I don't have those platforms at my disposal easily.

I've got both ia64 and ppc64 boxes in my cube. If ya have need (notice I didn't
say "want") to get on one, just ask... :) Otherwise, I'm happy to try out the
fix myself.

Created attachment 149606 [details]
Updated anaconda SRPM with loader patch.

Here is a patched SRPM for anaconda-11.2.0.31 that you can rebuild locally if
you need to.

Created attachment 149613 [details]
New initrd.img for IA64

New network boot initrd.img for IA64.  Please test.

Created attachment 149615 [details]
New ramdisk.image.gz for PPC64

New ramdisk.image.gz for PPC64.  Please test.

Please test these new ramdisk images with a rawhide-20070308 tree.

No luck, still bombed on ia64. Slightly different backtrace this time (did http
install, plus I presume code added may have changed line numbers):

loader received SIGSEGV!  Backtrace:                                           
[0x4000000000000530]
[0xa0000000000107e0]
[0x40000000000aabd0]
[0x40000000000aac40]
[0x4000000000031410]
[0x400000000003f0f0]
[0x400000000002fc70]
[0x4000000000030ce0]
[0x4000000000002dd0]
[0x4000000000005450]
install exited abnormally [1/1] 
sending termination signals...done
sending kill signals...done
disabling swap...
unmounting filesystems...
        /proc/bus/usb done
        /proc done
        /dev/pts done
        /sys done
        /tmp/ramfs done
you may safely reboot your system

$ eu-addr2line -e loader.debug < backtrace 
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1272
??:0
/usr/src/debug/anaconda-11.2.0.31/isys/str.c:87
/usr/src/debug/anaconda-11.2.0.31/isys/str.c:107
/usr/src/debug/anaconda-11.2.0.31/loader2/net.c:1932
/usr/src/debug/anaconda-11.2.0.31/loader2/urls.c:179
/usr/src/debug/anaconda-11.2.0.31/loader2/urlinstall.c:60
/usr/src/debug/anaconda-11.2.0.31/loader2/urlinstall.c:114
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1150
/usr/src/debug/anaconda-11.2.0.31/loader2/loader.c:1574

Another string being NULL.  Fixing.  Going ahead and building an anaconda for
tonight's build, we'll go with tomorrow's tree and see if things work or are
still broken.

OK, I've been testing the fixes today and it looks like I've fixed all of the
things that SIGSEGV that have been reported here. (famous last words).

Closing as rawhide and building a new anaconda.