[Bug tdep/34519] RISC-V: GDB crashes when disassembling elf32-bigriscv produced by -mbig-endian, and remote PC is byte-swapped

https://sourceware.org/bugzilla/show_bug.cgi?id=34519

--- Comment #5 from liu xu <mumuxi_ll at outlook dot com> ---
(In reply to Tom Tromey from comment #4)
> (In reply to liu xu from comment #3)
> 
> > > Could you try the gdb 18 pre-release or git master?
> 
> > I have reproduced a RISC-V big-endian remote debugging register endian issue
> > with GDB 18.0.90.
> 
> > Offline GDB disassembly works correctly and GDB recognizes the file as
> > elf32-bigriscv.
> 
> Ok, it sounds like upgrading gdb fixed the originally-reported bug.
> 
> > This suggests the RISC-V big-endian remote register path is applying the
> > wrong byte order for register packets. 
> 
> Can you "set debug remote 1" before the "target remote"?
> This will show a log of the remote protocol traffic.
> Maybe from this we can see if the bug is in gdb or the remote.

Hi Tom,

I followed your suggestion and reran the test with `set debug remote 1`; the
remote log shows that the remaining issue is on the remote gdbstub side, where
RISC-V big-endian register packets are encoded with the wrong byte order.

For example, after hitting a breakpoint at `0x80000010`, the `$g` packet
contains `10000080` for PC and GDB reports `pc = 0x10000080`; after `set $a0 =
0x12345678`, the next `$g` packet contains `78563412` for `a0`.

After fixing the remote gdbstub register read path to use the runtime RISC-V
endian state, LE/BE GDB against LE/BE ELF all works correctly.

Thanks for your help.

-- 
You are receiving this mail because:
You are on the CC list for the bug.