[valgrind] [Bug 438630] Adding zero variants of arm64 v8.2 FCMEQ, FCMGE, FCMGT, FCMLE and FCMLT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 ahashmi changed: What|Removed |Added Status|RESOLVED|CLOSED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438630] Adding zero variants of arm64 v8.2 FCMEQ, FCMGE, FCMGT, FCMLE and FCMLT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 ahashmi changed: What|Removed |Added Status|REPORTED|RESOLVED Resolution|--- |FIXED --- Comment #3 from ahashmi --- Landed 393732dda164c1cc0fc511eadc0b8f06008ade4f -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038 ahashmi changed: What|Removed |Added Status|RESOLVED|CLOSED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038 ahashmi changed: What|Removed |Added Status|REPORTED|RESOLVED Resolution|--- |FIXED --- Comment #6 from ahashmi --- Landed c5331315d7a62ec1c117ff773fadfeb0bf433b02 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038
--- Comment #5 from ahashmi ---
(In reply to Julian Seward from comment #3)
> No .. ensure SIGILL is raised (decode failure) for any undecodable insns.
> Never assert.
Good point, I've changed it to:
+ else {
+/* ty = 2 is an illegal encoding */
+if (sigill_diag) {
+ vex_printf("ARM64 front end: dis_AdvSIMD_fp_conditional_compare\n");
+}
+return False;
+ }
> + return True;
>
> return False;
>
> This is a strange sequence. The second return is unreachable. (Maybe I
> misread?)
My bad. The second return is a leftover from some debug testing which I should
have deleted.
> As a nit, in such situations I'd prefer if you set these both to NULL so as to
> protect against future snafus .. I know it's redundant as the code stands.
> Viz:
>
> HReg (*iselExpr)(ISelEnv*, IRExpr*) = NULL;
> ARM64Instr* (*VCmp)(HReg, HReg) = NULL
Good point! initialised.
Bugs caused by uninitialised data can be insidious. The code is fine now but
possible future changes in the subsequent if clauses could allow such bugs to
appear.
--
You are receiving this mail because:
You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038 ahashmi changed: What|Removed |Added Attachment #139231|0 |1 is obsolete|| --- Comment #4 from ahashmi --- Created attachment 139733 --> https://bugs.kde.org/attachment.cgi?id=139733=edit Adds arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions Updated patch to address issues raised by review. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438630] Adding zero variants of arm64 v8.2 FCMEQ, FCMGE, FCMGT, FCMLE and FCMLT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 --- Comment #1 from ahashmi --- Created attachment 139381 --> https://bugs.kde.org/attachment.cgi?id=139381=edit Adds zero variants of arm64 v8.2 FCMEQ, FCMGE, FCMGT, FCMLE and FCMLT instructions NOTE: This patch is for the compare-with-zero variants of instructions currently at review in https://bugs.kde.org/show_bug.cgi?id=438038 and is based on a branch which includes those changes. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438630] Adding zero variants of arm64 v8.2 FCMEQ, FCMGE, FCMGT, FCMLE and FCMLT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 ahashmi changed: What|Removed |Added Summary|Adding zero variants of |Adding zero variants of |arm64 v8.2 FCMEQ, FCMGE and |arm64 v8.2 FCMEQ, FCMGE, |FCMGT instructions |FCMGT, FCMLE and FCMLT ||instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438630] Adding zero variants of arm64 v8.2 FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 ahashmi changed: What|Removed |Added Summary|Adding zero variants of |Adding zero variants of |arm64 v8.2 FCCMP, FCCMPE, |arm64 v8.2 FCMEQ, FCMGE and |FCMEQ, FCMGE and FCMGT |FCMGT instructions |instructions| -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #8 from ahashmi --- Adding compare-with-zero variants of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions: https://bugs.kde.org/show_bug.cgi?id=438630 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038 --- Comment #2 from ahashmi --- Compare-with-zero variants of these instructions are implemented by https://bugs.kde.org/show_bug.cgi?id=438630 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438630] Adding zero variants of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 ahashmi changed: What|Removed |Added Assignee|jsew...@acm.org |assad.has...@linaro.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438630] New: Adding zero variants of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438630 Bug ID: 438630 Summary: Adding zero variants of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- This patch is part of work adding arm64 v8.2 support as summarised in container bug https://bugs.kde.org/show_bug.cgi?id=428016 The register variant of this patch is https://bugs.kde.org/show_bug.cgi?id=438038 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038 ahashmi changed: What|Removed |Added Summary|Addition of arm64 v8.2 |Addition of arm64 v8.2 |FCCMP, FCCMPE and FCMEQ,|FCCMP, FCCMPE, FCMEQ, FCMGE |FCMGE and FCMGT |and FCMGT instructions |instructions| -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE and FCMEQ, FCMGE and FCMGT instructions
https://bugs.kde.org/show_bug.cgi?id=438038 ahashmi changed: What|Removed |Added Summary|Addition of arm64 v8.2 |Addition of arm64 v8.2 |FCCMP, FCCMPE and FCMEQ |FCCMP, FCCMPE and FCMEQ, |instructions|FCMGE and FCMGT ||instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE and FCMEQ instructions
https://bugs.kde.org/show_bug.cgi?id=438038 --- Comment #1 from ahashmi --- Created attachment 139231 --> https://bugs.kde.org/attachment.cgi?id=139231=edit Adds arm64 v8.2 FCCMP, FCCMPE, FCMEQ, FCMGE and FCMGT instructions Note to reviewers: this patch may look a bit big at first glance but there's a lot of boilerplate test code and fairly straightforward IR construction additions. Patch has been tested with --memcheck:leak-check=yes --tool=memcheck --track-origins=yes -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 ahashmi changed: What|Removed |Added Status|RESOLVED|CLOSED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 ahashmi changed: What|Removed |Added Resolution|--- |FIXED Status|REPORTED|RESOLVED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 --- Comment #6 from ahashmi --- Landed, f048559cb04bd88f0613d02a3b7aa5fdd075617b. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 --- Comment #5 from ahashmi --- (In reply to Julian Seward from comment #4) > I didn't actually see any gating stuff in the three new frontend > (guest_arm64_toIR.c) cases, but maybe it's done elsewhere? Yes, the three new front-end cases are in dis_AdvSIMD_three_same_fp16() all of which is gated because that function only handles half-precision instructions, VEX_HWCAPS_ARM64_FP16. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 --- Comment #3 from ahashmi --- (In reply to Julian Seward from comment #2) > This is structurally very similar to the scalar-version patch at > https://bugs.kde.org/show_bug.cgi?id=436411#c2 (which is good) and > so the same comments for that patch (in comment 3) apply here, > meaning, the concern about hwcaps-gating. The review fixes for the scalar variants patch https://bugs.kde.org/show_bug.cgi?id=436411 (hwcaps-gating and iselIntExpr_RIL_wrk issues) means this latest patch works. Tested with regtest and combinations of --memcheck:leak-check=yes --tool=memcheck --track-origins=yes. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Status|RESOLVED|CLOSED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Resolution|--- |FIXED Status|REPORTED|RESOLVED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 --- Comment #7 from ahashmi --- Landed, 3ac86151404e6d018473dc3f6cd8b933be1ee038. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Attachment #138947|0 |1 is obsolete|| --- Comment #6 from ahashmi --- Created attachment 139074 --> https://bugs.kde.org/attachment.cgi?id=139074=edit Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions > That all looks fine; please land. Just before you do -- if you haven't > already -- please though run the test cases once by hand on Memcheck, > with --track-origins=yes, and check nothing breaks. This has been > known to happen in the past, at least to me :-) Good point! And now I remember why I made the RIL change . When run with --tool=memcheck: vex: priv/host_arm64_isel.c:1292 (iselIntExpr_RIL_wrk): Assertion `ty == Ity_I64 || ty == Ity_I32' failed. Thanks for the out-of-band email explanation about IR traversal for isel and the RIL functions. The latest patch includes the fix and tests pass with --tool=memcheck and --track-origins=yes. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #7 from ahashmi --- Addition of arm64 v8.2 FCCMP, FCCMPE and FCMEQ instructions: https://bugs.kde.org/show_bug.cgi?id=438038 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] Addition of arm64 v8.2 FCCMP, FCCMPE and FCMEQ instructions
https://bugs.kde.org/show_bug.cgi?id=438038 ahashmi changed: What|Removed |Added Assignee|jsew...@acm.org |assad.has...@linaro.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 438038] New: Addition of arm64 v8.2 FCCMP, FCCMPE and FCMEQ instructions
https://bugs.kde.org/show_bug.cgi?id=438038 Bug ID: 438038 Summary: Addition of arm64 v8.2 FCCMP, FCCMPE and FCMEQ instructions Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- This patch is part of work adding arm64 v8.2 support as summarised in container bug https://bugs.kde.org/show_bug.cgi?id=428016 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Attachment #138343|0 |1 is obsolete|| --- Comment #4 from ahashmi --- Created attachment 138947 --> https://bugs.kde.org/attachment.cgi?id=138947=edit Adds arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions Thanks for reviewing, apologies for the delay in responding. The first issue pertaining to CPU feature-gating of the front-end was something I did implement in the first v8.2 instruction, FADDP (https://bugs.kde.org/show_bug.cgi?id=413547) but for some reason ommitted in later v8.2 work! I've added the VEX_HWCAPS_ARM64_FP16 check for all half-precision FP instructions. The second issue to do with the Iex_Const case for iselIntExpr_RIL_wrk() turns out to be some sort of confusion on my part! I can't remember exactly what happened but I think an assert to catch Ity_I16 must've fired partway through development and I just assumed it was the the Iex_Const in iselIntExpr_RIL_wrk(). Anyway, I've removed the unnecessary changes and tested on v8.x where x<3. I'd prefer to get this patch past review and merged before addressing the same hwcaps-gating issue in https://bugs.kde.org/show_bug.cgi?id=436873. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 --- Comment #1 from ahashmi --- Created attachment 138348 --> https://bugs.kde.org/attachment.cgi?id=138348=edit Adds arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions Adds arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions NOTE: This patch is for the vector instruction variant and is based on a branch which includes the scalar patch currently at review in https://bugs.kde.org/show_bug.cgi?id=436411 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #6 from ahashmi --- Addition of vector FABD, FACGE, FACGT and FADD instructions: https://bugs.kde.org/show_bug.cgi?id=436873 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Attachment #138216|0 |1 is obsolete|| CC||assad.has...@linaro.org --- Comment #2 from ahashmi --- Created attachment 138343 --> https://bugs.kde.org/attachment.cgi?id=138343=edit Adds arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions This is an updated patch which fixes a bug I found while working on the vector variants of these instructions. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 ahashmi changed: What|Removed |Added Assignee|jsew...@acm.org |assad.has...@linaro.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436873] New: Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436873 Bug ID: 436873 Summary: Addition of arm64 v8.2 vector FABD, FACGE, FACGT and FADD instructions Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- This patch is part of work adding arm64 v8.2 support as summarised in container bug https://bugs.kde.org/show_bug.cgi?id=428016 The scalar variant of this patch is https://bugs.kde.org/show_bug.cgi?id=436411 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #5 from ahashmi --- Addittion of v8.2 scalar FABD, FACGE, FACGT and FADD instructions: https://bugs.kde.org/show_bug.cgi?id=436411 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 --- Comment #1 from ahashmi --- Created attachment 138216 --> https://bugs.kde.org/attachment.cgi?id=138216=edit Adds arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions This patch is for the scalar instruction variant, the vector patch will follow next. I have left the 16 bit immediate handling in iselIntExpr_RIL_wrk() as a TODO while I get to grips with v8.2 immediate encodings. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 scalar FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Summary|Addition of arm64 v8.2 |Addition of arm64 v8.2 |FABD, FACGE, FACGT and FADD |scalar FABD, FACGE, FACGT |instructions|and FADD instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] Addition of arm64 v8.2 FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 ahashmi changed: What|Removed |Added Assignee|jsew...@acm.org |assad.has...@linaro.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 436411] New: Addition of arm64 v8.2 FABD, FACGE, FACGT and FADD instructions
https://bugs.kde.org/show_bug.cgi?id=436411 Bug ID: 436411 Summary: Addition of arm64 v8.2 FABD, FACGE, FACGT and FADD instructions Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- This patch is part of work adding arm64 v8.2 support as summarised in container bug https://bugs.kde.org/show_bug.cgi?id=428016 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 ahashmi changed: What|Removed |Added Status|RESOLVED|CLOSED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 ahashmi changed: What|Removed |Added Resolution|--- |FIXED Status|REPORTED|RESOLVED --- Comment #7 from ahashmi --- Landed, 7593a4773997807695d6514e9d01a60cb489851c -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 --- Comment #5 from ahashmi --- > This isn't quite right, in that the two added lines change the behaviour for > the fallthrough cases immediately above. Gah, of course! Careless of me. Fixed. > One final question: did you check that the relevant test programs run OK > (without any assertion failures) on memcheck, by running them by hand, with > --tool=memcheck and also --tool=memcheck --track-origins=yes ? Yes, there are no assertions and nothing breaks. Alse checked debug output displays new mnemonics and operands (using --trace-flags=1000 --tool=none). -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 ahashmi changed: What|Removed |Added Attachment #135211|0 |1 is obsolete|| --- Comment #4 from ahashmi --- Created attachment 135237 --> https://bugs.kde.org/attachment.cgi?id=135237=edit Adds arm64 v8.2 FABS, FNEG and FSQRT instructions Fixes bug raised in review (case Iop_Sqrt16Fx8: position). -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 --- Comment #2 from ahashmi --- When ready for checkin, please paste the following text as the Git commit message: - - - snip 432161 Addition of arm64 v8.2 FADDP, FNEG and FSQRT This patch adds FP half-precision support for the following: FABS , FABS ., . FNEG , FNEG ., . FSQRT , FSQRT ., . Fixes https://bugs.kde.org/show_bug.cgi?id=432161 - - - snip Unless there are any suggestions for improvements, I'll try and keep this Git commit message format for all future arm64 v8.2 instruction patches. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 --- Comment #1 from ahashmi --- Created attachment 135211 --> https://bugs.kde.org/attachment.cgi?id=135211=edit Adds arm64 v8.2 FABS, FNEG and FSQRT instructions Patch ready for review. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #4 from ahashmi --- Addition of FABS, FNEG and FSQRT instructions: https://bugs.kde.org/show_bug.cgi?id=432161 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 432161] New: Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions
https://bugs.kde.org/show_bug.cgi?id=432161 Bug ID: 432161 Summary: Addition of arm64 v8.2 FABS, FNEG and FSQRT instructions Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- Addition of arm64 half-precision v8.2 FABS, FNEG and FSQRT instructions. This is part of arm64 v8.2 container bug https://bugs.kde.org/show_bug.cgi?id=428016 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 431556] Complete arm64 FADDP v8.2 instruction support started in 413547
https://bugs.kde.org/show_bug.cgi?id=431556 ahashmi changed: What|Removed |Added Attachment #134852|0 |1 is obsolete|| --- Comment #5 from ahashmi --- Created attachment 134859 --> https://bugs.kde.org/attachment.cgi?id=134859=edit Patch completes addition of arm64 v8.2 faddp instructions. Changes after review. Removed redundant; - comment in math_REARRANGE_FOR_FLOATING_PAIRWISE() - N_N_N test case variants in fp_and_simd_v82.c -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 431556] Complete arm64 FADDP v8.2 instruction support started in 413547
https://bugs.kde.org/show_bug.cgi?id=431556 --- Comment #4 from ahashmi --- (In reply to Julian Seward from comment #3) > The cases are distinguished as follows: > isD == True, bitQ == 1 => 2d > isD == False, bitQ == 1 => 4s > isD == False, bitQ == 0 => 2s > + isH == True, bitQ == 0 => 4h > + isH == False, bitQ == 1 => 8h > > Is this comment out of date? The function it applies to takes an > ARM64VecESize now, > not isH / isD. Ah yes! It is out-of-date. I'll remove. > + if (1) test_faddp_4h_00_00_00(TyH); > > The tests where the three register numbers are the same .. are they of any > value? In particular, they won't expose mixups where the wrong register > number is used in decode. Those cases are covered by the _N_N+1_N+2 variants > afaics. Is there some other reason to keep the N_N_N variants? The reason for the N_N_N variants is that they may flush out bugs to do with temporaries in IR. That's my simple minded motivation. If such bugs are not possible I can remove. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 431556] Complete arm64 FADDP v8.2 instruction support started in 413547
https://bugs.kde.org/show_bug.cgi?id=431556 ahashmi changed: What|Removed |Added Attachment #134815|0 |1 is obsolete|| --- Comment #2 from ahashmi --- Created attachment 134852 --> https://bugs.kde.org/attachment.cgi?id=134852=edit Patch completes addition of arm64 v8.2 faddp instructions. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #3 from ahashmi --- Addition of v8.2 faddp instructions: https://bugs.kde.org/show_bug.cgi?id=431556 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 431556] Complete arm64 FADDP v8.2 instruction support started in 413547
https://bugs.kde.org/show_bug.cgi?id=431556 --- Comment #1 from ahashmi --- Created attachment 134815 --> https://bugs.kde.org/attachment.cgi?id=134815=edit arm64 v8.2 faddp instruction support Patch completes addition of arm64 v8.2 faddp instructions. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 431556] Complete arm64 FADDP v8.2 instruction support started in 413547
https://bugs.kde.org/show_bug.cgi?id=431556 ahashmi changed: What|Removed |Added Assignee|jsew...@acm.org |assad.has...@linaro.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 431556] New: Complete arm64 FADDP v8.2 instruction support started in 413547
https://bugs.kde.org/show_bug.cgi?id=431556 Bug ID: 431556 Summary: Complete arm64 FADDP v8.2 instruction support started in 413547 Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- SUMMARY Complete arm64 FADDP v8.2 instruction support started in 413547. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 430484] none/tests/arm64/atomics_v81 occassionally fails
https://bugs.kde.org/show_bug.cgi?id=430484 --- Comment #1 from ahashmi --- CPU implementer 0x43 and CPU part 0x0af indicates a Cavium ThunderX2. Previous experience with DynamoRIO suggests this behaviour may be limited to ThunderX machines only. It would be interesting to see if the same occasional failures occur on non-ThunderX machines. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 --- Comment #9 from ahashmi --- > Most of the code in this patch is, I assume, boilerplate test-support > functions taken from other tests in none/tests/arm64. Yes but it includes additional code for testing half-precision floating-point instructions. See comment at Float16 definition. > Should this be present? It certainly looks like it shouldn't be; if it > accepts the instruction then it gives no implementation for it, which is > definitely bogus. Added implementation and tests for scalar version of half-precision FADDP. The hope is that this can be the template for future patches supporting v8.2 instructions. > "most" is somewhat misleading; "half" might be more accurate. Ditto in > randFloat. Done. > Just as a general sanity check .. is the way that the capabilities are > established (by looking at auxv HWCAPS) consistent with the outcome of the > discussion in bug 414268 ? Yes. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 ahashmi changed: What|Removed |Added Attachment #133751|0 |1 is obsolete|| --- Comment #8 from ahashmi --- Created attachment 134154 --> https://bugs.kde.org/attachment.cgi?id=134154=edit Adds arm64 feature detection for regression tests and v8.2 faddp Patch update after first set of review comments (on 2020-12-09 10:10:23 UTC). -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 --- Comment #17 from ahashmi --- > One tiny nit .. please #undef get_cpu_ftr and get_ftr just before the > closing brace of the fn that defines them. Good spot! Fixed in latest patch. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added Attachment #133750|0 |1 is obsolete|| --- Comment #16 from ahashmi --- Created attachment 133936 --> https://bugs.kde.org/attachment.cgi?id=133936=edit Adds arm64 feature detection for v8.x instructions #undef get_cpu_ftr and get_ftr after use. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 --- Comment #6 from ahashmi --- > Assad, am I right to understand that the comment 3 patch > has been made redundant by the comment 4 patch? That's correct Julian. I should have deleted the comment 3 patch when I uploaded the latest comment 4 patch. Can't seem to delete it after the comment 4 patch upload. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 --- Comment #13 from ahashmi --- > So it seems to me we also need a way to filter the attribute/feature > registers to only advertise those armv8 instructions we actually implement. > That is what other arches do too (see for example the dirtyhelper_CPUID in > guest_x86_helpers.c or the s390x_dirtyhelper_vec_op). ISWYM now, thanks for shining a light on an aspect of Valgrind's transparency I hadn't thought about. I would prefer to implement such a guest feature support view/filter as a separate patch. I've created https://bugs.kde.org/show_bug.cgi?id=430117 to track this change. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 430117] New: Implement guest Arm64/AArch64 feature detection support for v8 instructions
https://bugs.kde.org/show_bug.cgi?id=430117 Bug ID: 430117 Summary: Implement guest Arm64/AArch64 feature detection support for v8 instructions Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- SUMMARY The bug https://bugs.kde.org/show_bug.cgi?id=414268 added host AArch64 feature detection and decoding for v8.x. However, Valgrind must provide a view of v8.x instructions supported by Valgrind to the guest which will not be the same as those supported by the host, until they have been implemented in Valgrind. ADDITIONAL INFORMATION A helper function needs to be implemented which provides a filtered view to the guest of a set of instructions implemented by Valgrind. This is not the same as the set of instructions supported by the host which Valgrind reads in order to decode instructions which may or may not be supported by Valgrind. When instructions are decoded, this helper function is planted in the IR to provide the guest with the filtered view of what Valgrind wants the guest to think are the instructions supported by the host. As an existing example, see the x86 implementation of populating guest state with SSE1 support: x86g_dirtyhelper_CPUID_sse1(). This sets VexGuestX86State with SSE1 features supported by Valgrind. A pointer to the function is planted in IR by unsafeIRDirty_0_N() when SSE1 instructions are decoded. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 --- Comment #11 from ahashmi --- > I know too little about arm64 to know whether or not we should care about > these. The current approach assumes that all base v8 h/w has the features you list. I've not seen any base v8 h/w which is different from that list. AFAICT: - We don't need to make any changes to Valgrind's existing base v8 support in terms of establishing and checking features. - Time has shown that no bugs have been raised to do with lack of base v8 feature support. > Since programs can also check features through the attribute/feature > registers, do we also need to filter those when inspected by the guest code? Hmm...I don't see why we should. We aim to faithfully support features detected, so if guest code reads those it should expect Valgrind to support it. If Valgrind doesn't, it means we haven't yet implemented support for that feature. (If I understand your question correctly.) -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 --- Comment #4 from ahashmi --- Created attachment 133751 --> https://bugs.kde.org/attachment.cgi?id=133751=edit Adds arm64 feature detection for regression tests Removed v8.x version detection. Now using features only. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 --- Comment #9 from ahashmi --- > It's true that if you are v8.2 or better then DC CVAP is present, but this > does not mean the converse (that if DC CVAP is present then you are v8.2). It > is valid for a v8.1 implementation to provide this feature. That makes a difference to my decision. I made an incorrect assumption when reading the spec: - - - snip 0b DC CVAP not supported. 0b0001 DC CVAP supported. 0b0010 DC CVAP and DC CVADP supported. ... In Armv8.2, the permitted values are 0b0001 and 0b0010. >From Armv8.5, the only permitted value is 0b0010 - - - snip "In Armv8.2" != "From Armv8.2" But some features are more tightly bound to a version, e.g. RDM - - - snip 0b No RDMA instructions implemented. 0b0001 SQRDMLAH and SQRDMLSH instructions implemented. All other values are reserved. >From Armv8.1, the only permitted value is 0b0001. - - - snip So, Arm publishes a version specification with a set of mandatory and optional features. An implementer builds h/w which states conformance to a version. The implementer's version conformance statement is fairly meaningless without a list of features which are supported. This answers the question: "why doesn't /proc/cpuinfo give an Arm version?" -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added Attachment #133666|0 |1 is obsolete|| --- Comment #8 from ahashmi --- Created attachment 133750 --> https://bugs.kde.org/attachment.cgi?id=133750=edit Adds arm64 feature detection for v8.x instructions Removed v8.x version detection/setting. Now using feature list only. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 ahashmi changed: What|Removed |Added Attachment #133311|0 |1 is obsolete|| --- Comment #3 from ahashmi --- Created attachment 133667 --> https://bugs.kde.org/attachment.cgi?id=133667=edit Adds arm64 feature detection for regression tests -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268
--- Comment #6 from ahashmi ---
Thanks for reviewing/testing Mark Wielaard and Peter Maydell.
> There are two machines this seems to work correctly on:
> Arch and hwcaps: ARM64, LittleEndian, v8.1-rdm-atomics
> Arch and hwcaps: ARM64, LittleEndian, v8.2-dpcvap-rdm-atomics-fp16-vfp16
>
> But one where I get Arch and hwcaps: ARM64, LittleEndian, v8.1 and then the
> impossible happens:
>
> VEX: SQRDML{AH|SH} instructions are mandatory from v8.1 onwards.
> Found: v8.1
> Cannot continue. Good-bye
Fixed.
Just goes to show I need to test on a wider range of hardware before
submitting!
> The right way to test fields in these ID registers is to look at the entire
> 4-bit number and test that it is greater than or equal to the value
> representing the feature you're looking for. (See D13.1.3 "Principles of the
> ID
> scheme for fields in ID registers" in the v8A Arm ARM.)
Fixed. Good spot, it was staring me in the face!
> For A-profile my recommendation is as far as you possibly can to avoid having
> any "is this v8.1?" or "is this v8.2?" tests and instead to stick strictly to
> "is feature X present?". The architecture permits a v8.x implementation to
I thought about this disconnect between version and supported feature set when
implementing the initial patch and considered omitting versions altogether. I
decided against it because I don't yet have enough experience with Valgrind,
particularly the decode-to-IR logic and the tools pipeline. On the basis that
the version is set when its mandatory option(s) are established and it may
simplify things down the line, I prefer to leave this for now. If it transpires
versions are redundant, I'll remove them at the earliest sign.
--
You are receiving this mail because:
You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added Attachment #133310|0 |1 is obsolete|| --- Comment #5 from ahashmi --- Created attachment 133666 --> https://bugs.kde.org/attachment.cgi?id=133666=edit Patch adds arm64 feature detection for v8.x instructions Patch update after first set of review comments. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 ahashmi changed: What|Removed |Added CC||jsew...@acm.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added CC||jsew...@acm.org -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added See Also||https://bugs.kde.org/show_b ||ug.cgi?id=413547 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 ahashmi changed: What|Removed |Added See Also||https://bugs.kde.org/show_b ||ug.cgi?id=414268 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 ahashmi changed: What|Removed |Added Attachment #123538|0 |1 is obsolete|| --- Comment #2 from ahashmi --- Created attachment 133311 --> https://bugs.kde.org/attachment.cgi?id=133311=edit Patch adds arm64 feature detection for regression tests This patch implements arm64 ISA features/attributes checks for regression tests, and must be committed after https://bugs.kde.org/show_bug.cgi?id=414268 The simple test of a v8.2 encoding shows how the fix for https://bugs.kde.org/show_bug.cgi?id=414268 can be used in IR creation on different arm64 versions. When both patches are committed, 'make regtest' should result in the following for v8.0 and 8.1: - - - snip fp_and_simd: valgrind -q ./fp_and_simd fp_and_simd_v82: (skipping, prereq failed: test -x fp_and_simd_v82 && ../../../tests/arm64_features fphp) integer: valgrind -q ./integer - - - snip The test should run and pass on v8.2: - - - snip fp_and_simd: valgrind -q ./fp_and_simd fp_and_simd_v82: valgrind -q ./fp_and_simd_v82 integer: valgrind -q ./integer - - - snip -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added CC||assad.has...@linaro.org --- Comment #2 from ahashmi --- Created attachment 133310 --> https://bugs.kde.org/attachment.cgi?id=133310=edit Patch adds arm64 feature detection for v8.x instructions This patch adds hwcaps support for arm64. It uses the arm recommended method of hwcaps feature detection where a reliable auxilliary vector's AT_HWCAP may not be available. Instruction set attribute and feature registers are read to establish the v8 architecture level as well as optional and mandatory ISA features. This patch can be tested with the patch which fixes https://bugs.kde.org/show_bug.cgi?id=413547 This patch must be committed before https://bugs.kde.org/show_bug.cgi?id=413547 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 413547] regression test does not check for Arm 64 features
https://bugs.kde.org/show_bug.cgi?id=413547 ahashmi changed: What|Removed |Added Blocks|414270 | Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=414270 [Bug 414270] Collection of bugs for new ARMv8.1 work and features -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414270] Collection of bugs for new ARMv8.1 work and features
https://bugs.kde.org/show_bug.cgi?id=414270 ahashmi changed: What|Removed |Added Resolution|--- |FIXED Status|REPORTED|RESOLVED -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0)
https://bugs.kde.org/show_bug.cgi?id=414268 ahashmi changed: What|Removed |Added Blocks|414270 | Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=414270 [Bug 414270] Collection of bugs for new ARMv8.1 work and features -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414270] Collection of bugs for new ARMv8.1 work and features
https://bugs.kde.org/show_bug.cgi?id=414270 ahashmi changed: What|Removed |Added Depends on|413547, 414268 | Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=413547 [Bug 413547] regression test does not check for Arm 64 features https://bugs.kde.org/show_bug.cgi?id=414268 [Bug 414268] Enable AArch64 feature detection and decoding for v8.x instructions (where x>0) -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414270] Collection of bugs for new ARMv8.1 work and features
https://bugs.kde.org/show_bug.cgi?id=414270 --- Comment #1 from ahashmi --- Feature detection tasks moved to https://bugs.kde.org/show_bug.cgi?id=428016 (v8.2): https://bugs.kde.org/show_bug.cgi?id=414268 https://bugs.kde.org/show_bug.cgi?id=413547 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #2 from ahashmi --- Feature detection tasks moved from https://bugs.kde.org/show_bug.cgi?id=414270 (v8.1) and made part of v8.2 work. https://bugs.kde.org/show_bug.cgi?id=414268 https://bugs.kde.org/show_bug.cgi?id=413547 -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 --- Comment #1 from ahashmi --- Created attachment 132579 --> https://bugs.kde.org/attachment.cgi?id=132579=edit v8.2 instruction list (including variants) -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 428016] New: Collection of tasks for addition of ARMv8.2 instructions and related features
https://bugs.kde.org/show_bug.cgi?id=428016 Bug ID: 428016 Summary: Collection of tasks for addition of ARMv8.2 instructions and related features Product: valgrind Version: unspecified Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- Created attachment 132577 --> https://bugs.kde.org/attachment.cgi?id=132577=edit v8.2 instruction list (excluding variants) SUMMARY This container is for a collection of bugs to add Arm v8.2 instruction support. STEPS TO REPRODUCE Compile subject source with -march=armv8.2-a. If the compiler generates any of the instructions in the attached list (v8.2_excl_variants.txt), valgrind will exit with unknown instruction error. ADDITIONAL INFORMATION There are 71 v8.2 instructions, see attached list v8.2_excl_variants.txt. Most of these have scalar, vector, paired, fixed and floating point variants which brings the total number of instructions which need to be implemented to 123, see attached list v8.2_incl_variants.txt. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 421570] std_mutex fails on Arm v8.1 h/w
https://bugs.kde.org/show_bug.cgi?id=421570 --- Comment #2 from ahashmi --- Improved test coverage for the Arm v8.1 instructions will be added in a later patch. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 421570] std_mutex fails on Arm v8.1 h/w
https://bugs.kde.org/show_bug.cgi?id=421570 --- Comment #1 from ahashmi --- Created attachment 128478 --> https://bugs.kde.org/attachment.cgi?id=128478=edit Enable v8.1 atomics and fix SWP and LDUMAX instrs -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 421570] New: std_mutex fails on Arm v8.1 h/w
https://bugs.kde.org/show_bug.cgi?id=421570 Bug ID: 421570 Summary: std_mutex fails on Arm v8.1 h/w Product: valgrind Version: 3.14.0 Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- SUMMARY The atomics test drd/tests/std_mutex hangs on Arm v8.1 h/w when built with GCC10. STEPS TO REPRODUCE 1. make 2. make regtest OBSERVED RESULT std_mutex hangs EXPECTED RESULT SOFTWARE/OS VERSIONS Windows: macOS: Linux/KDE Plasma: (available in About System) KDE Plasma Version: KDE Frameworks Version: Qt Version: ADDITIONAL INFORMATION -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 421321] gcc10 arm64 build needs __getauxval for linking with libgcc
https://bugs.kde.org/show_bug.cgi?id=421321 ahashmi changed: What|Removed |Added CC||assad.has...@linaro.org --- Comment #3 from ahashmi --- Created attachment 128471 --> https://bugs.kde.org/attachment.cgi?id=128471=edit Fix patch for std_mutex spin bug -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 418702] ARMv8.1 Paired register compare-and-swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=418702 ahashmi changed: What|Removed |Added Attachment #126711|0 |1 is obsolete|| --- Comment #3 from ahashmi --- Created attachment 126714 --> https://bugs.kde.org/attachment.cgi?id=126714=edit Patch tests ARM v8.1 CASP instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 418702] ARMv8.1 Paired register compare-and-swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=418702 --- Comment #2 from ahashmi --- Created attachment 126711 --> https://bugs.kde.org/attachment.cgi?id=126711=edit Patch tests ARM v8.1 CASP instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 418702] ARMv8.1 Paired register compare-and-swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=418702 --- Comment #1 from ahashmi --- Created attachment 126710 --> https://bugs.kde.org/attachment.cgi?id=126710=edit Patch implements ARM v8.1 CASP instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 418702] ARMv8.1 Paired register compare-and-swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=418702 ahashmi changed: What|Removed |Added Blocks||414270 Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=414270 [Bug 414270] Collection of bugs for new ARMv8.1 work and features -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414270] Collection of bugs for new ARMv8.1 work and features
https://bugs.kde.org/show_bug.cgi?id=414270 ahashmi changed: What|Removed |Added Depends on||418702 Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=418702 [Bug 418702] ARMv8.1 Paired register compare-and-swap instructions are not supported -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 418702] New: ARMv8.1 Paired register compare-and-swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=418702 Bug ID: 418702 Summary: ARMv8.1 Paired register compare-and-swap instructions are not supported Product: valgrind Version: 3.14.0 Platform: Other OS: Linux Status: REPORTED Severity: normal Priority: NOR Component: vex Assignee: jsew...@acm.org Reporter: assad.has...@linaro.org Target Milestone: --- SUMMARY The following paired register compare-and-swap instructions are not currently supported: CASP, CASPA, CASPAL, CASPL STEPS TO REPRODUCE Compile with -march=armv8.1-a and if the compiler generates any of the above, valgrind will exit with unknown instruction error. ADDITIONAL INFORMATION These are also referred to as double register compare-and-swap instructions on some architectures. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=415136 ahashmi changed: What|Removed |Added Attachment #126174|0 |1 is obsolete|| --- Comment #6 from ahashmi --- Created attachment 126176 --> https://bugs.kde.org/attachment.cgi?id=126176=edit Patch implements ARM v8.1 CAS instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=415136 --- Comment #5 from ahashmi --- Thanks for reviewing in detail. Good point about splitting up the large patch. I've uploaded the implementation and tests as 2 separate patches: arm64_v81_cas_impl.patch arm64_v81_cas_test.patch > I am going to need some context to make sense of this... The loop in this context is not looping for the same reason guest code using a CAS instruction would loop. In this case the cbne w8, loop instruction is checking to see if the previous store exclusive worked. The spec says that store exclusives return 0 in the w register if the store updated memory, 1 if it failed to update. This way we always make sure that memory has been updated before returning from the CAS. Previously it would have returned even if the store failed. > Also, what is the reason for changing the hardwired register numbers? Adding the store check loop means we need to compare the store's status value in a w register and since w8 is the scratch register we use that, leaving w1/x1 untouched after the load. -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=415136 ahashmi changed: What|Removed |Added Attachment #124590|0 |1 is obsolete|| --- Comment #3 from ahashmi --- Created attachment 126174 --> https://bugs.kde.org/attachment.cgi?id=126174=edit Patch implements ARM v8.1 CAS instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=415136 --- Comment #4 from ahashmi --- Created attachment 126175 --> https://bugs.kde.org/attachment.cgi?id=126175=edit Patch tests ARM v8.1 CAS instructions -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=415136 --- Comment #1 from ahashmi --- Created attachment 124590 --> https://bugs.kde.org/attachment.cgi?id=124590=edit Patch implements and tests ARM v8.1 CAS instructions This patch adds support for the following ARM v8.1 instructions: CAS, CASA, CASAL, CASL CASB, CASAB, CASALB, CASLB CASH, CASAH, CASALH, CASLH (Note this patch does not add support for the pairwise CAS instructions: CASP, CASPA, CASPAL, CASPL) -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 414270] Collection of bugs for new ARMv8.1 work and features
https://bugs.kde.org/show_bug.cgi?id=414270 ahashmi changed: What|Removed |Added Depends on||415136 Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=415136 [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported -- You are receiving this mail because: You are watching all bug changes.
[valgrind] [Bug 415136] ARMv8.1 Compare-and-Swap instructions are not supported
https://bugs.kde.org/show_bug.cgi?id=415136 ahashmi changed: What|Removed |Added Blocks||414270 Referenced Bugs: https://bugs.kde.org/show_bug.cgi?id=414270 [Bug 414270] Collection of bugs for new ARMv8.1 work and features -- You are receiving this mail because: You are watching all bug changes.
