[REVIEW][PATCH 0/3] signal/ia64: siginfo fixes and cleanups

[REVIEW][PATCH 0/3] signal/ia64: siginfo fixes and cleanups

[Eric W. Biederman]

I have been slowly cleaning up the architectues ever since I discovered
that the pattern of passing in struct siginfo is error prone, and
occassionally results in borken siginfo being sent to userspace.

It turns out that ia64 is one of the noticable offenders.  Ever since
ia64 was merged in 2.3.43 it has been generating incorrect siginfo for
SIGSEGV assocaited with the inability to setup a signal stack frame
or cleanup a signal stack frame.

This is pretty esoteric case and the problem was only some values that
should be 0 being something else so I am not surpised it was missed.

The following series fixes the two buggy cases I have found and then
cleans up replaces force_sig_info with force_sig_fault to avoid errors
of this kind in the future.

I don't think I have made made any mistakes in this coversion but if
people can look the code over and see if they can spot anything I would
appreciate it.

My intention is to merge this through my siginfo tree.  If you feel it
should go through your arch tree let me know.   All of the prerequisites
should have been merged several releases ago.

Eric W. Biederman (3):
      signal/ia64: Use the generic force_sigsegv in setup_frame
      signal/ia64: Use the force_sig(SIGSEGV,...) in ia64_rt_sigreturn
      signal/ia64: Use force_sig_fault where appropriate

 arch/ia64/kernel/brl_emu.c   |  31 ++--------
 arch/ia64/kernel/signal.c    |  60 ++++--------------
 arch/ia64/kernel/traps.c     | 144 ++++++++++++-------------------------------
 arch/ia64/kernel/unaligned.c |  12 +---
 arch/ia64/mm/fault.c         |  12 +---
 5 files changed, 62 insertions(+), 197 deletions(-)

Eric

[REVIEW][PATCH 0/3] signal/ia64: siginfo fixes and cleanups

[Eric W. Biederman]

I have been slowly cleaning up the architectues ever since I discovered
that the pattern of passing in struct siginfo is error prone, and
occassionally results in borken siginfo being sent to userspace.

It turns out that ia64 is one of the noticable offenders.  Ever since
ia64 was merged in 2.3.43 it has been generating incorrect siginfo for
SIGSEGV assocaited with the inability to setup a signal stack frame
or cleanup a signal stack frame.

This is pretty esoteric case and the problem was only some values that
should be 0 being something else so I am not surpised it was missed.

The following series fixes the two buggy cases I have found and then
cleans up replaces force_sig_info with force_sig_fault to avoid errors
of this kind in the future.

I don't think I have made made any mistakes in this coversion but if
people can look the code over and see if they can spot anything I would
appreciate it.

My intention is to merge this through my siginfo tree.  If you feel it
should go through your arch tree let me know.   All of the prerequisites
should have been merged several releases ago.

Eric W. Biederman (3):
      signal/ia64: Use the generic force_sigsegv in setup_frame
      signal/ia64: Use the force_sig(SIGSEGV,...) in ia64_rt_sigreturn
      signal/ia64: Use force_sig_fault where appropriate

 arch/ia64/kernel/brl_emu.c   |  31 ++--------
 arch/ia64/kernel/signal.c    |  60 ++++--------------
 arch/ia64/kernel/traps.c     | 144 ++++++++++++-------------------------------
 arch/ia64/kernel/unaligned.c |  12 +---
 arch/ia64/mm/fault.c         |  12 +---
 5 files changed, 62 insertions(+), 197 deletions(-)

Eric

Re: [REVIEW][PATCH 0/3] signal/ia64: siginfo fixes and cleanups

[Luck, Tony]

On Mon, Sep 24, 2018 at 04:20:32PM +0200, Eric W. Biederman wrote:
> 
> It turns out that ia64 is one of the noticable offenders.  Ever since
> ia64 was merged in 2.3.43 it has been generating incorrect siginfo for
> SIGSEGV assocaited with the inability to setup a signal stack frame
> or cleanup a signal stack frame.

I'll blame that on the "previous administration". I didn't start work
on Linux until 2.3.99, and didn't take over as maintainer until 2.6.11 :-)

> I don't think I have made made any mistakes in this coversion but if
> people can look the code over and see if they can spot anything I would
> appreciate it.

I don't see any mistakes either. It builds and boots. Obviously I don't
have a test case for this issue (or it would have been found long ago).

> My intention is to merge this through my siginfo tree.  If you feel it
> should go through your arch tree let me know.   All of the prerequisites
> should have been merged several releases ago.

Sure. Merge away.

Acked-by: Tony Luck <tony.luck@intel.com>

-Tony

Re: [REVIEW][PATCH 0/3] signal/ia64: siginfo fixes and cleanups

[Luck, Tony]

On Mon, Sep 24, 2018 at 04:20:32PM +0200, Eric W. Biederman wrote:
> 
> It turns out that ia64 is one of the noticable offenders.  Ever since
> ia64 was merged in 2.3.43 it has been generating incorrect siginfo for
> SIGSEGV assocaited with the inability to setup a signal stack frame
> or cleanup a signal stack frame.

I'll blame that on the "previous administration". I didn't start work
on Linux until 2.3.99, and didn't take over as maintainer until 2.6.11 :-)

> I don't think I have made made any mistakes in this coversion but if
> people can look the code over and see if they can spot anything I would
> appreciate it.

I don't see any mistakes either. It builds and boots. Obviously I don't
have a test case for this issue (or it would have been found long ago).

> My intention is to merge this through my siginfo tree.  If you feel it
> should go through your arch tree let me know.   All of the prerequisites
> should have been merged several releases ago.

Sure. Merge away.

Acked-by: Tony Luck <tony.luck@intel.com>

-Tony