Hi,
> > As of revision 2762, the v2 repository is still BROKEN and superiotool
> > doesn't appear in v3 either (checked revision 486).
> >
> > Such commits are one reason why I think self-acked commits should not be
> > allowed.
>
> Well. Half-hearted code reviews are an illusion of safety, too.
>
> The current code's lar handling is pretty much broken even though that
> code was reviewed, but obviously nobody even tested it:
>
> Wrote LinuxBIOS table at: 0x00000500 - 0x00000a54 checksum 401f
> Show all devs...
> cpus: Unknown device path type: 0
> Stage2 code done.
> LAR: Attempting to open 'normal/payload'.
> LAR: Attempting to open 'normal/payload/segment0'.
> LAR: load_file: No such file 'normal/payload/segment0'
>
> It just hangs somewhere bogus and never gets to the defined error in
>
> arch/x86/stage1.c:194: die("FATAL: No usable payload found.\n");
>
> also the current code contains quite some warnings again regarding
> int-to-pointer casts in lar.c and pointer-to-int cast and uninitialized
> variable use in stage1.c
run_address gets called with entry=0.
Attached patch fixes it.
(If thread hi-jacking is forbidden, sorry)
--
Alex
Index: arch/x86/stage1.c
===================================================================
--- arch/x86/stage1.c (revision 486)
+++ arch/x86/stage1.c (working copy)
@@ -189,7 +189,8 @@
entry = newentry;
}
printk(BIOS_SPEW, "all loaded, entry %p\n", entry);
- run_address(entry);
+ if (entry)
+ run_address(entry);
die("FATAL: No usable payload found.\n");
--
linuxbios mailing list
[email protected]
http://www.linuxbios.org/mailman/listinfo/linuxbios
