Vedant Kumar via llvm-dev
2019-Nov-15 18:40 UTC
[llvm-dev] Commit history duplicated, seeing weird diffusion activity (Was: [Diffusion] rG67c416dc9a5a: [DebugInfo] Allow spill slots in call site parameter descriptions)
I just got a Diffusion notification about a change of mine being reverted by
Fangrui, but I'm not sure that's actually what happened and am confused
and concerned.
My commit was "[DebugInfo] Allow spill slots in call site parameter
descriptions", and it appears in the history under two hashes: 1ee84e and
67c416. The first commit contains the actual change. The second touches
something in yaml2obj, which I promise I did not do:
```
$ g show 67c416dc9a5aeda034d46bdcb3b63fffa462d28e
commit 67c416dc9a5aeda034d46bdcb3b63fffa462d28e
Author: Vedant Kumar <vsk at apple.com>
Date: Thu Nov 14 09:20:58 2019
[DebugInfo] Allow spill slots in call site parameter descriptions
[snip]
diff --git a/llvm/test/tools/yaml2obj/coff-xrelocs.yaml
b/llvm/test/tools/yaml2obj/COFF/xrelocs.yaml
similarity index 100%
rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml
rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml
```
Does anyone know what happened?
thanks,
vedant
> On Nov 15, 2019, at 10:06 AM, Fangrui Song via Phabricator <reviews at
reviews.llvm.org> wrote:
>
> MaskRay committed rG67c416dc9a5a: [DebugInfo] Allow spill slots in call
site parameter descriptions (authored by vsk).
> MaskRay added a reverted change: rGeb4c98ca3d25: [DebugInfo] Exclude memory
location values as parameter entry values.
>
> [DebugInfo] Allow spill slots in call site parameter descriptions
>
> Allow call site paramter descriptions to reference spill slots. Spill
> slots are not visible to high-level LLVM IR, so they can safely be
> referenced during entry value evaluation (as they cannot be clobbered by
> some other function).
>
> This gives a 5% increase in the number of call site parameter DIEs in an
> LTO x86_64 build of the xnu kernel.
>
> This reverts commit eb4c98ca3d2590bad9f6542afbf3a7824d2b53fa (
> [DebugInfo] Exclude memory location values as parameter entry values),
> effectively reintroducing the portion of D60716
<https://reviews.llvm.org/D60716> which dealt with memory
> locations (authored by Djordje, Nikola, Ananth, and Ivan).
>
> This partially addresses llvm.org/PR43343. However, not all memory
> operands forwarded to callees live in spill slots. In the xnu build, it
> may be possible to use an escape analysis to increase the number of call
> site parameter by another 15% (more details in PR43343).
>
> Differential Revision: https://reviews.llvm.org/D70254
>
>
> BRANCHES
> master
>
> Files:
> /llvm/test/tools/yaml2obj/COFF/
> /llvm/test/tools/yaml2obj/COFF/xrelocs.yaml
> /llvm/test/tools/yaml2obj/coff-xrelocs.yaml
>
> PATCH
>
> diff --git a/llvm/test/tools/yaml2obj/coff-xrelocs.yaml
b/llvm/test/tools/yaml2obj/COFF/xrelocs.yaml
> similarity index 100%
> rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml
> rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml
>
> Users:
> vsk (Author)
> MaskRay (Committer)
>
> https://reviews.llvm.org/rG67c416dc9a5a
>
>
>
Fāng-ruì Sòng via llvm-dev
2019-Nov-15 18:51 UTC
[llvm-dev] Commit history duplicated, seeing weird diffusion activity (Was: [Diffusion] rG67c416dc9a5a: [DebugInfo] Allow spill slots in call site parameter descriptions)
I am also very confused about rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml It may be due to my misoperation. git checkout master; git merge --ff objcopy-redefine-sym && git push # to push the commit "[llvm-objcopy][MachO] Implement --redefine-sym and --redefine-syms" I intended to append the worktree change to another branch yaml2obj-move but probably accidentally amended master. Sorry for causing the confusion. On Fri, Nov 15, 2019 at 10:40 AM Vedant Kumar <vedant_kumar at apple.com> wrote:> > I just got a Diffusion notification about a change of mine being reverted by Fangrui, but I'm not sure that's actually what happened and am confused and concerned. > > My commit was "[DebugInfo] Allow spill slots in call site parameter descriptions", and it appears in the history under two hashes: 1ee84e and 67c416. The first commit contains the actual change. The second touches something in yaml2obj, which I promise I did not do: > > ``` > $ g show 67c416dc9a5aeda034d46bdcb3b63fffa462d28e > commit 67c416dc9a5aeda034d46bdcb3b63fffa462d28e > Author: Vedant Kumar <vsk at apple.com> > Date: Thu Nov 14 09:20:58 2019 > > [DebugInfo] Allow spill slots in call site parameter descriptions > > [snip] > > diff --git a/llvm/test/tools/yaml2obj/coff-xrelocs.yaml b/llvm/test/tools/yaml2obj/COFF/xrelocs.yaml > similarity index 100% > rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml > rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml > ``` > > Does anyone know what happened? > > thanks, > vedant > > > On Nov 15, 2019, at 10:06 AM, Fangrui Song via Phabricator <reviews at reviews.llvm.org> wrote: > > > > MaskRay committed rG67c416dc9a5a: [DebugInfo] Allow spill slots in call site parameter descriptions (authored by vsk). > > MaskRay added a reverted change: rGeb4c98ca3d25: [DebugInfo] Exclude memory location values as parameter entry values. > > > > [DebugInfo] Allow spill slots in call site parameter descriptions > > > > Allow call site paramter descriptions to reference spill slots. Spill > > slots are not visible to high-level LLVM IR, so they can safely be > > referenced during entry value evaluation (as they cannot be clobbered by > > some other function). > > > > This gives a 5% increase in the number of call site parameter DIEs in an > > LTO x86_64 build of the xnu kernel. > > > > This reverts commit eb4c98ca3d2590bad9f6542afbf3a7824d2b53fa ( > > [DebugInfo] Exclude memory location values as parameter entry values), > > effectively reintroducing the portion of D60716 <https://reviews.llvm.org/D60716> which dealt with memory > > locations (authored by Djordje, Nikola, Ananth, and Ivan). > > > > This partially addresses llvm.org/PR43343. However, not all memory > > operands forwarded to callees live in spill slots. In the xnu build, it > > may be possible to use an escape analysis to increase the number of call > > site parameter by another 15% (more details in PR43343). > > > > Differential Revision: https://reviews.llvm.org/D70254 > > > > > > BRANCHES > > master > > > > Files: > > /llvm/test/tools/yaml2obj/COFF/ > > /llvm/test/tools/yaml2obj/COFF/xrelocs.yaml > > /llvm/test/tools/yaml2obj/coff-xrelocs.yaml > > > > PATCH > > > > diff --git a/llvm/test/tools/yaml2obj/coff-xrelocs.yaml b/llvm/test/tools/yaml2obj/COFF/xrelocs.yaml > > similarity index 100% > > rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml > > rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml > > > > Users: > > vsk (Author) > > MaskRay (Committer) > > > > https://reviews.llvm.org/rG67c416dc9a5a > > > > > > >-- 宋方睿
Vedant Kumar via llvm-dev
2019-Nov-15 20:00 UTC
[llvm-dev] Commit history duplicated, seeing weird diffusion activity (Was: [Diffusion] rG67c416dc9a5a: [DebugInfo] Allow spill slots in call site parameter descriptions)
Ahh ok, no problem. Still not sure I understand exactly how the fast-forward duplicated the message from 1ee84e, but it doesn't look like there's any issue with the repo or tooling, so my main concern is assuaged. thanks vedant> On Nov 15, 2019, at 10:51 AM, Fāng-ruì Sòng <maskray at google.com> wrote: > > I am also very confused about > > rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml > rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml > > > It may be due to my misoperation. > > git checkout master; git merge --ff objcopy-redefine-sym && git push > # to push the commit "[llvm-objcopy][MachO] Implement --redefine-sym > and --redefine-syms" > I intended to append the worktree change to another branch > yaml2obj-move but probably accidentally amended master. > Sorry for causing the confusion. > > On Fri, Nov 15, 2019 at 10:40 AM Vedant Kumar <vedant_kumar at apple.com> wrote: >> >> I just got a Diffusion notification about a change of mine being reverted by Fangrui, but I'm not sure that's actually what happened and am confused and concerned. >> >> My commit was "[DebugInfo] Allow spill slots in call site parameter descriptions", and it appears in the history under two hashes: 1ee84e and 67c416. The first commit contains the actual change. The second touches something in yaml2obj, which I promise I did not do: >> >> ``` >> $ g show 67c416dc9a5aeda034d46bdcb3b63fffa462d28e >> commit 67c416dc9a5aeda034d46bdcb3b63fffa462d28e >> Author: Vedant Kumar <vsk at apple.com> >> Date: Thu Nov 14 09:20:58 2019 >> >> [DebugInfo] Allow spill slots in call site parameter descriptions >> >> [snip] >> >> diff --git a/llvm/test/tools/yaml2obj/coff-xrelocs.yaml b/llvm/test/tools/yaml2obj/COFF/xrelocs.yaml >> similarity index 100% >> rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml >> rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml >> ``` >> >> Does anyone know what happened? >> >> thanks, >> vedant >> >>> On Nov 15, 2019, at 10:06 AM, Fangrui Song via Phabricator <reviews at reviews.llvm.org> wrote: >>> >>> MaskRay committed rG67c416dc9a5a: [DebugInfo] Allow spill slots in call site parameter descriptions (authored by vsk). >>> MaskRay added a reverted change: rGeb4c98ca3d25: [DebugInfo] Exclude memory location values as parameter entry values. >>> >>> [DebugInfo] Allow spill slots in call site parameter descriptions >>> >>> Allow call site paramter descriptions to reference spill slots. Spill >>> slots are not visible to high-level LLVM IR, so they can safely be >>> referenced during entry value evaluation (as they cannot be clobbered by >>> some other function). >>> >>> This gives a 5% increase in the number of call site parameter DIEs in an >>> LTO x86_64 build of the xnu kernel. >>> >>> This reverts commit eb4c98ca3d2590bad9f6542afbf3a7824d2b53fa ( >>> [DebugInfo] Exclude memory location values as parameter entry values), >>> effectively reintroducing the portion of D60716 <https://reviews.llvm.org/D60716> which dealt with memory >>> locations (authored by Djordje, Nikola, Ananth, and Ivan). >>> >>> This partially addresses llvm.org/PR43343. However, not all memory >>> operands forwarded to callees live in spill slots. In the xnu build, it >>> may be possible to use an escape analysis to increase the number of call >>> site parameter by another 15% (more details in PR43343). >>> >>> Differential Revision: https://reviews.llvm.org/D70254 >>> >>> >>> BRANCHES >>> master >>> >>> Files: >>> /llvm/test/tools/yaml2obj/COFF/ >>> /llvm/test/tools/yaml2obj/COFF/xrelocs.yaml >>> /llvm/test/tools/yaml2obj/coff-xrelocs.yaml >>> >>> PATCH >>> >>> diff --git a/llvm/test/tools/yaml2obj/coff-xrelocs.yaml b/llvm/test/tools/yaml2obj/COFF/xrelocs.yaml >>> similarity index 100% >>> rename from llvm/test/tools/yaml2obj/coff-xrelocs.yaml >>> rename to llvm/test/tools/yaml2obj/COFF/xrelocs.yaml >>> >>> Users: >>> vsk (Author) >>> MaskRay (Committer) >>> >>> https://reviews.llvm.org/rG67c416dc9a5a >>> >>> >>> >> > > > -- > 宋方睿