Re: Segfault in libfontconfig
Matt Dainty <[email protected]>
| Newsgroups | gmane.os.netbsd.ports.hpcsh,gmane.os.netbsd.ports.dreamcast |
|---|---|
| Message-ID | <[email protected]> |
On Tue, 2004-09-14 at 22:45, Valeriy E. Ushakov wrote: > On Tue, Sep 14, 2004 at 22:37:10 +0400, Valeriy E. Ushakov wrote: > > > > fontconfig library uses functions that return structures. SuperH ABI > > > uses r2 to pass the pointer to the memory for the returned structure. > > > But dynamic linker happily blows r2 away (it's a scratch register > > > otherwise) while doing the dynamic symbol resolution. > > > > I wonder if _rtld_bind_start (libexec/ld.elf_so/arch/sh3/rtld_start.S) > > should save/restore r2 instead of r3. > > I have filed port-sh3/26949 and committed a fix for it, revision 1.4 > of libexec/ld.elf_so/arch/sh3/rtld_start.S > > I will request the pull-up to 2.0, but I really NEED YOUR HELP. I > need someone to test this on the 2.0 branch (2.0 is at revision 1.3 of > that file, so pulling this fix up is trivial). The 2.0 freeze is > already in effect, so I can't request the pull-up w/out a clean build > + boot test on at least one sh3 platform (well, in theory all pull-ups > should be tested like this, freeze or not :). But I have absolutely > no time to do that, so I'd very much appreciate if someone can do the > test and drop me a mail. What do I need to do? Just rebuild the whole tree again? Sadly I've not been masochistic enough to try it on the Jornada itself, but I can have the whole thing built in about 30 minutes or less on my i386 rig, if that's acceptable. > PS: The original bug should be reproducible with fc-list - which > doesn't require a running X server. Yup, I forgot to mention I triggered it this way too. Matt