This is the mail archive of the
gdb@sources.redhat.com
mailing list for the GDB project.
Question for ARM experts
- From: Kris Warkentin <kewarken at qnx dot com>
- To: gdb at sources dot redhat dot com
- Date: Wed, 30 Jun 2004 14:55:13 -0400
- Subject: Question for ARM experts
In arm-tdep.c I see that if we get GDB_OSABI_UNKNOWN we assume
GDB_OSABI_ARM_APCS. What would the consequences of not doing that be
for other arm targets?
The problem is, there isn't anything particularily distinct about QNX
Neutrino binaries so I've got a generic Neutrino sniffer that basically
says, "Am I connected to a Neutrino target? Okay, then let's use the
Neutrino ABI stuff." Arm is the only one that falls apart for this
because when I return GDB_OSABI_QNXNTO and arm_elf_osabi_sniffer() in
arm-tdep.c returns APCS, gdbarch_lookup_osabi() goes, "Argh!! Two
OSABIs...*choke*"
So here's the thing: if I allow arm_elf_osabi_sniffer to return
GDB_OSABI_UNKNOWN in the case of ELFOSABI_ARM, my sniffer works fine.
The question is, if we allow the APCS fallback in the case of
ELFOSABI_UNKNOWN but not in ELFOSABI_ARM, is that going to break the world?
If so, does anyone have any suggestions for how to deal with this?
cheers,
Kris