Re: [pcre-dev] [Bug 2365] New: check_wordboundary does not …

Top Page

Reply to this message
Author: zatlas1@yahoo.com
Date:  
To: admin@bugs.exim.org, pcre-dev@exim.org, mike.munday@ibm.com
Subject: Re: [pcre-dev] [Bug 2365] New: check_wordboundary does not explicitly set zero flag
Hello MikeAre you aware that there is a current implementation of the non-jit version for IBM Z series.  I obviously support your efforts to do a jit version.  I am the maintainer of that stuff and would like to work with you so I couldA.  Share with you my experienceB. Coordinate between the two versions
I see that you are an IBM employee, so your designs will take precedence over mine, but with coordination we could keep them close.
I am glad that IBM decided to take over.
Please contact me at zatlas1@???
Ze'ev Atlas

Sent from Yahoo Mail on Android

On Wed, Jan 16, 2019 at 10:27 AM, admin@???<admin@???> wrote: https://bugs.exim.org/show_bug.cgi?id=2365

            Bug ID: 2365
          Summary: check_wordboundary does not explicitly set zero flag
          Product: PCRE
          Version: N/A
          Hardware: Other
                OS: All
            Status: NEW
          Severity: bug
          Priority: medium
        Component: Code
          Assignee: ph10@???
          Reporter: mike.munday@???
                CC: pcre-dev@???

Created attachment 1166
  --> https://bugs.exim.org/attachment.cgi?id=1166&action=edit
Fix

The OP_WORD_BOUNDARY/OP_NOT_WORD_BOUNDARY implementation calls the fast
function check_wordboundary expecting it to set the zero flag (SLJIT_SET_Z).
The function never explicitly sets the flag though. I presume this works on
x86_64 because the xor instruction sets the zero flag regardless of whether it
is requested or not.

I've attached a patch that fixes the issue for me on the architecture I am
trying to port the PCRE2 JIT to (IBM Z).

--
You are receiving this mail because:
You are on the CC list for the bug.
--
## List details at https://lists.exim.org/mailman/listinfo/pcre-dev