[PATCH v4 00/11] Split up s390-linux-tdep.c

[Philipp Rudo <prudo at linux dot vnet dot ibm dot com>]

Hi everybody

changes v3 -> v4:

	- Rebase to current master.

	- In patch #7 move init of OSABI as late as possible in gdbarch_init
	  in order for the OSABI to be able to overwrite the defaults.

	- Split up the split up patch, i.e. move the record-replay feature
	  in a separate patch.  This is simply due to the fact that the
	  combined patch is too large for the mailing list.  This requires
	  a small hack, in particular the process_record must be set in
	  init_osabi shortly to keep the set bisectable.  The hack is
	  cleaned up again when record-replay gets moved.

Note: I can no longer reproduce the problem in py-breakpoint.exp I reported
(https://sourceware.org/ml/gdb-patches/2017-12/msg00383.html).  That's why I
haven't had a closer look at it.  For me it looks like the problem solved
itself.  Thanks for whoever did this.

changes v2 -> v3:

	- Rebase to current master.

	- Fix bug Yao reported in Patch #1 (s390: Remove duplicate checks
	  for cached gdbarch at init).

	- Move s390_guess_tracepoint_registers, s390_stap_is_single_operand,
	  s390_gcc_target_options and, s390_gnu_triplet_regexp to common
	  s390-tdep.

	- Move s390_process_record and its helper functions to common
	  s390-tdep.  This requires new patch #8.

	- Move s390_frame_unwind, s390_stub_frame_unwind, s390_frame_base
	  and their helpers to common s390-tdep

	- Clean up s390-tdep.h, i.e. remove unneeded exports.

	- Squash former patch #11 (s390: Add comments to uncommented
	  functions in s390-linux-tdep.c) into patch #10 (s390: Clean up
	  s390-linux-tdep.c).  The new code moves left the clean up patch
	  nearly empty and both patches are more the less cleanups.

	- Squash former patch #10 (s390: Add comments to uncommented
	  functions in s390-tdep.c) into patch #9 (s390: Split up
	  s390-linux-tdep.c into two files).  The extra patch isn't really
	  needed so reduce the number of patches.

	- Fix white space damages in s390_process_record and its helpers.

changes v1 -> v2:

	- Rebase to current master.

	- Hook s390 into osabi mechanism including some preparation/clean
	  up in s390_gdbarch_init (Patches #1-#7)

	- Don't move s390_cannot_store_register and s390_write_pc as well as
	  the regmap/regset definitions to the new file.

Thanks
Philipp

---

This patch set splits up the s390 tdep code followed by some minor clean up
and coding style fixes.  It originates from my Linux kernel feature and
helps adding the new Linux kernel 'OS' to s390 while keeping the code
manageable.  I want to bring the set upstream in advance of the rest of the
kernel debugging feature as nearly all changes to s390-linux-tdep.c cause
merge conflicts making the maintenance off-tree extremely labor intensive
and error prone.

Thanks
Philipp


Philipp Rudo (11):
  s390: Remove duplicate checks for cached gdbarch at init
  s390: Allocate gdbarch & tdep at start of gdbarch init
  s390: gdbarch_tdep.have_* int -> bool
  s390: gdbarch_tdep add field tdesc
  s390: Move tdesc validation to separate function
  s390: if -> gdb_assert for tdesc_has_registers check
  s390: Hook s390 into OSABI mechanism
  s390: gdbarch_tdep add hook for syscall record
  s390: Split up s390-linux-tdep.c into two files
  s390: Move record-replay to s390-tdep
  s390: Clean up s390-linux-tdep.c

 gdb/Makefile.in       |    3 +
 gdb/configure.tgt     |    4 +-
 gdb/s390-linux-nat.c  |    1 +
 gdb/s390-linux-tdep.c | 7226 +------------------------------------------------
 gdb/s390-linux-tdep.h |  176 +-
 gdb/s390-tdep.c       | 7124 ++++++++++++++++++++++++++++++++++++++++++++++++
 gdb/s390-tdep.h       |  318 +++
 7 files changed, 7549 insertions(+), 7303 deletions(-)
 create mode 100644 gdb/s390-tdep.c
 create mode 100644 gdb/s390-tdep.h

-- 
2.13.5