diff options
| author | Marcin KoĆcielnicki <koriakin@0x04.net> | 2016-03-12 14:03:26 +0100 | 
|---|---|---|
| committer | Marcin KoĆcielnicki <koriakin@0x04.net> | 2016-03-30 01:51:06 +0200 | 
| commit | 28170b88cc8b40fdea2b065dafe6e1872a47ee4e (patch) | |
| tree | cb05b89e30b424c98961d3e3eb13e63f29055484 /gdb/python/python-internal.h | |
| parent | a08b52b5c45195c0b095215f19422d2ab67a3a8d (diff) | |
| download | binutils-28170b88cc8b40fdea2b065dafe6e1872a47ee4e.zip binutils-28170b88cc8b40fdea2b065dafe6e1872a47ee4e.tar.gz binutils-28170b88cc8b40fdea2b065dafe6e1872a47ee4e.tar.bz2  | |
gdbserver: Handle 'v' packet while processing qSymbol.
On powerpc64, qSymbol query may require gdb to read a function
descriptor, sending a vFile packet to gdbserver.  Thus, we need
to handle 'v' packet in look_up_one_symbol.
vFile replies may be quite long, and require reallocating own_buf.
Since handle_v_requests assumes the buffer is the static global own_buf
from server.c and reallocates it, we need to make own_buf global and
use it from look_up_one_symbol instead of using our own auto variable.
I've also done the same change in relocate_instruction, just in case.
On gdb side, in remote_check_symbols, rs->buf may be clobbered by vFile
handling, yet we need its contents for the reply (the symbol name is
stored there).  Allocate a new buffer instead.
This broke fast tracepoints on powerpc64, due to errors in reading IPA
symbols.
gdb/ChangeLog:
	* remote.c (remote_check_symbols): Allocate own buffer for reply.
gdbserver/ChangeLog:
	* remote-utils.c (look_up_one_symbol): Remove own_buf, handle 'v'
	packets.
	(relocate_instruction): Remove own_buf.
	* server.c (own_buf): Make global.
	(handle_v_requests): Make global.
	* server.h (own_buf): New declaration.
	(handle_v_requests): New prototype.
Diffstat (limited to 'gdb/python/python-internal.h')
0 files changed, 0 insertions, 0 deletions
