This is the mail archive of the
mailing list for the GDB project.
[rfc][0/13] Eliminate read_register et al
- From: "Ulrich Weigand" <uweigand at de dot ibm dot com>
- To: gdb-patches at sourceware dot org
- Date: Thu, 7 Jun 2007 22:56:38 +0200 (CEST)
- Subject: [rfc][0/13] Eliminate read_register et al
the following patch set eliminates the routines read_register,
read_register_pid, write_register, and write_register_pid, including
all their callers. This is the next step towards removing the
current_regcache global variable.
The patch set consists of 13 parts:
First some special read_register et al uses:
[1/13] write_register in solib-sunos.c
[2/13] read_register in cris/frv_stopped_data_address
[3/13] read_register in mep-tdep.c current_ routines
[4/13] read_register in deprecated_mips_set_processor_regs_hack
[5/13] read_register in sh_show_regs_command
[6/13] read_register (invalidly) used in prologue analyzers
Then more general classes of uses:
[7/13] Uses that are already unnecessary
[8/13] Uses in software_single_step callbacks
[9/13] Uses in skip_trampoline callbacks
[10/13] Uses in get_longjmp_target callbacks
[11/13] Uses in push_dummy_code callbacks
[12/13] Uses in read_pc / write_pc callbacks
[13/13] Remove definition
I'd appreciate any comments; in particular some the areas touched by
the first set of patches could probably need a more general cleanup
I've tested the full patch set natively on:
and on the simulator for:
Architectures touched but untested as I have neither native
hardware nor a working (toolchain or) simulator:
Additional testers would certainly be welcome!
In addition, tested by running gdb-mbuild.sh and making sure every
file touched by the patch set still builds.
Any opinions? Should I proceed with this patch set?
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE