- User Since
- Jun 6 2014, 3:56 PM (272 w, 14 h)
Mar 21 2019
Mar 15 2019
Mar 4 2019
Jan 17 2019
Dec 28 2018
Dec 17 2018
Would you like to abandon this in favor of "Transition to C11 memory model" Trac #16044?
Dec 9 2018
Inline comments done.
- Refactor PPC remainder code
Dec 8 2018
I plan the following
Dec 5 2018
What can I do to move this forward?
Nov 27 2018
Actually, the former was done as part of D5300.
CI keeps failing, now on Linux and OS X. I validated locally on powerpc64le and x86_64.
Nov 25 2018
- Break up long line
Nov 24 2018
This validates now on x86_64, powerpc64le needs D5300.
Nov 23 2018
I could not validate this because HEAD does not build for me (neither x86_64 nor powerpc64le). Perhaps I need to wait for things to be fixed in HEAD and rebase again.
Nov 18 2018
Nov 11 2018
Nov 5 2018
Nov 3 2018
Oct 31 2018
I guess I can abandon this. Thanks @alpmestan for your review and discussion of the issue!
I think this is fixed in HEAD now with commit 82a7164
Oct 29 2018
Oct 28 2018
All ‘goto‘s are inside the ‘#if defined x86_64...‘ so on other platforms none of the ‘goto‘s will be seen by the compiler because the preprocessor removed that code already.
What kind of problems did you see?
Oct 8 2018
I was hoping the fix would also help with Trac #15411 but it doesn't. So I removed the reference to that ticket.
Oct 7 2018
Aug 21 2018
Perhaps we can avoid negation and say what the guarantee of the barrier actually is.
Aug 3 2018
My comment was meant as a contribution of an idea to the discussion. Sorry, I should have made that explicit.
Aug 1 2018
Linux documentation of memory barriers defines a write barrier as follows:
Jul 25 2018
Jul 22 2018
Jun 10 2018
- Expand tabs.
May 26 2018
Mar 9 2018
Note: I don't have a working SPARC machine so I could not test this.
Mar 5 2018
I created D4468 for this.
Feb 1 2018
I think atomic read is fine but atomic write is missing a parameter. See inline comment.
Currently MO_SS_Conv W32 W64 is not supported on 32 bit PPC. In fact, we would get a compiler panic if I understand the code in nativeGen/PPC/Codegen.hs right.
Nov 27 2017
I have a small comment regarding the C functions.
Nov 14 2017
LGTM. I validated this diff applied to HEAD and Phab:D4181 on a POWER8 (powerpc64-linux) with no regressions.
Nov 12 2017
I built the diff against HEAD on a POWER8 and got this error:
Nov 6 2017
Nov 5 2017
- Improve comments
- Fix duplicate labels on powerpc64le
Nov 4 2017
I'd like to fix powerpc64le, too.
- Break up long lines
- Make block info tables local and simplify
- Simplify and always use infoTblLbl
- Add note Proc-point block entry-point
Nov 3 2017
Nov 2 2017
Oct 2 2017
Sep 30 2017
- add lost commit
- add comment for memory barrier
Sep 27 2017
For atomic read the answer is yes.
Sep 23 2017
I checked the POWER ISA docs and compared the code with what gcc does. I also added a comment with references to the POWER docs in a place where the code looks very odd.
- add comments and fix typo
Sep 22 2017
Sep 17 2017
May 9 2017
Let's use Ben's patch D3560. It is both cleaner and more efficient.
Yes. That is better than my patch.
May 5 2017
Apr 30 2017
Apr 25 2017
Mar 27 2017
- Fix count trailing zeros formula in comment
Mar 25 2017
Mar 24 2017
Fix long lines.
Rebase again to fix Habormaster builds.
Mar 21 2017
Rebase so it applies cleanly to HEAD again.
Jan 17 2017
Jan 16 2017
Jan 14 2017
- Break up long lines
Dec 18 2016
- remove newline.
Dec 16 2016
- Fix typos in comment.
- Improve RTS linker support detection
- RTS linker: Better error message on PPC64
- RTS linker: Note runtime-linker-support