cmd/compile: use proved bounds to remove signed division fix-ups #25239
Labels
FrozenDueToAge
help wanted
NeedsFix
The path to resolution is known, but the work has not been done.
Performance
Milestone
Comments
josharian
added
Performance
help wanted
NeedsFix
|
I would like to have a look at this, hoping to make ssa a little less opaque to me than it is now. |
|
Change https://golang.org/cl/130215 mentions this issue: |
|
Change https://golang.org/cl/131318 mentions this issue: |
|
Change https://golang.org/cl/131795 mentions this issue: |
|
Changes 131318 nd 131795 were sent by mistake - they have been abandoned. |
|
@rasky, @randall77, @josharian |
|
No you haven’t. Feel free to ping in the CL. As a general note, always
comment in Gerrit after you’re finished uploading a new patch, at least to
say PTAL; better would be to go through each open comment and close them
with “Done” or explaining why you haven’t done it. For instance, after my
comment and Keith’s, there’s no reply from you.
Il giorno dom 14 ott 2018 alle 05:54 Chris Liles <notifications@github.com>
ha scritto:
@rasky <https://github.com/rasky>, @randall77
<https://github.com/randall77>, @josharian <https://github.com/josharian>
Have I done something to stop CL 130215 in it's tracks?
Are you waiting for me to do something/fix something?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#25239 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AA95XdPG2FipYrZ16WMRnbpRco3Ptl9Tks5ukrV2gaJpZM4TxrKa>
.
--
Giovanni Bajo :: rasky@develer.com
Develer S.r.l. :: http://www.develer.com
Tel: +39-055-3984627 (ext: 206)
My Blog: http://giovanni.bajo.it
|
|
Thanks for that. I’m afraid the Gerrit process is still largely a mystery
to me. I thought I had replied to everything.
I will revisit the CL tomorrow and try to get it up to date.
I wanted to check because I can never seem to tell whether I should have
done something or not.
Again, thank you.
…On Sun, 14 Oct 2018 at 8:38 pm, Giovanni Bajo ***@***.***> wrote:
No you haven’t. Feel free to ping in the CL. As a general note, always
comment in Gerrit after you’re finished uploading a new patch, at least to
say PTAL; better would be to go through each open comment and close them
with “Done” or explaining why you haven’t done it. For instance, after my
comment and Keith’s, there’s no reply from you.
Il giorno dom 14 ott 2018 alle 05:54 Chris Liles ***@***.***
>
ha scritto:
> @rasky <https://github.com/rasky>, @randall77
> <https://github.com/randall77>, @josharian <https://github.com/josharian
>
> Have I done something to stop CL 130215 in it's tracks?
> Are you waiting for me to do something/fix something?
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#25239 (comment)>, or
mute
> the thread
> <
https://github.com/notifications/unsubscribe-auth/AA95XdPG2FipYrZ16WMRnbpRco3Ptl9Tks5ukrV2gaJpZM4TxrKa
>
> .
>
--
Giovanni Bajo :: ***@***.***
Develer S.r.l. :: http://www.develer.com
Tel: +39-055-3984627 (ext: 206)
My Blog: http://giovanni.bajo.it
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#25239 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AeohTkQJvLRRGCytCwhY3it9FVp4k10vks5ukwYogaJpZM4TxrKa>
.
|
|
There's some explanation here:
https://golang.org/doc/contribute.html#review
and I've got a pending CL to improve that a little bit (
https://go-review.googlesource.com/c/go/+/103575). See if you can spot what
you were missing, and/or feel free to suggest improvements to the wording.
In fact, new contributors' feedback is by far the most welcome for that
page, so don't be shy if something wasn't obvious.
Il giorno dom 14 ott 2018 alle ore 11:55 Chris Liles <
notifications@github.com> ha scritto:
… Thanks for that. I’m afraid the Gerrit process is still largely a mystery
to me. I thought I had replied to everything.
I will revisit the CL tomorrow and try to get it up to date.
I wanted to check because I can never seem to tell whether I should have
done something or not.
Again, thank you.
On Sun, 14 Oct 2018 at 8:38 pm, Giovanni Bajo ***@***.***>
wrote:
> No you haven’t. Feel free to ping in the CL. As a general note, always
> comment in Gerrit after you’re finished uploading a new patch, at least
to
> say PTAL; better would be to go through each open comment and close them
> with “Done” or explaining why you haven’t done it. For instance, after my
> comment and Keith’s, there’s no reply from you.
>
> Il giorno dom 14 ott 2018 alle 05:54 Chris Liles <
***@***.***
> >
> ha scritto:
>
> > @rasky <https://github.com/rasky>, @randall77
> > <https://github.com/randall77>, @josharian <
https://github.com/josharian
> >
> > Have I done something to stop CL 130215 in it's tracks?
> > Are you waiting for me to do something/fix something?
> >
> > —
> > You are receiving this because you were mentioned.
> > Reply to this email directly, view it on GitHub
> > <#25239 (comment)>, or
> mute
> > the thread
> > <
>
https://github.com/notifications/unsubscribe-auth/AA95XdPG2FipYrZ16WMRnbpRco3Ptl9Tks5ukrV2gaJpZM4TxrKa
> >
> > .
> >
> --
> Giovanni Bajo :: ***@***.***
> Develer S.r.l. :: http://www.develer.com
>
> Tel: +39-055-3984627 (ext: 206)
>
> My Blog: http://giovanni.bajo.it
>
> —
> You are receiving this because you commented.
> Reply to this email directly, view it on GitHub
> <#25239 (comment)>, or
mute
> the thread
> <
https://github.com/notifications/unsubscribe-auth/AeohTkQJvLRRGCytCwhY3it9FVp4k10vks5ukwYogaJpZM4TxrKa
>
> .
>
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#25239 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AA95XQnUJGRSz---5myWIxzfxgxpzlcZks5ukwoOgaJpZM4TxrKa>
.
--
Giovanni Bajo :: rasky@develer.com
Develer S.r.l. :: http://www.develer.com
Tel: +39-055-3984627 (ext: 206)
My Blog: http://giovanni.bajo.it
|
|
The best clue for me is lines 831-834 of your change.
I think I have been “replying” to the individual comments but I did not
realise that the replies are just batched up and not sent until I go back
and hit the Reply button.
In an earlier change I was driving Josharian crazy by only using the Reply
button and not responding to the comments individually. I thought I was on
the right track for this change. I am the world’s worst user of unfamiliar
software, I rarely figure out what to do by intuition.
Ah well, I’ll keep trying.
Anyway, your efforts have not been in vain - I am sure that you are correct
in thinking that a little bit of Gerrit information is needed in the
Guidelines document.
Thanks again
On Sun, 14 Oct 2018 at 9:17 pm, Giovanni Bajo <notifications@github.com>
wrote:
… There's some explanation here:
https://golang.org/doc/contribute.html#review
and I've got a pending CL to improve that a little bit (
https://go-review.googlesource.com/c/go/+/103575). See if you can spot
what
you were missing, and/or feel free to suggest improvements to the wording.
In fact, new contributors' feedback is by far the most welcome for that
page, so don't be shy if something wasn't obvious.
Il giorno dom 14 ott 2018 alle ore 11:55 Chris Liles <
***@***.***> ha scritto:
> Thanks for that. I’m afraid the Gerrit process is still largely a mystery
> to me. I thought I had replied to everything.
> I will revisit the CL tomorrow and try to get it up to date.
> I wanted to check because I can never seem to tell whether I should have
> done something or not.
> Again, thank you.
>
> On Sun, 14 Oct 2018 at 8:38 pm, Giovanni Bajo ***@***.***>
> wrote:
>
> > No you haven’t. Feel free to ping in the CL. As a general note, always
> > comment in Gerrit after you’re finished uploading a new patch, at least
> to
> > say PTAL; better would be to go through each open comment and close
them
> > with “Done” or explaining why you haven’t done it. For instance, after
my
> > comment and Keith’s, there’s no reply from you.
> >
> > Il giorno dom 14 ott 2018 alle 05:54 Chris Liles <
> ***@***.***
> > >
> > ha scritto:
> >
> > > @rasky <https://github.com/rasky>, @randall77
> > > <https://github.com/randall77>, @josharian <
> https://github.com/josharian
> > >
> > > Have I done something to stop CL 130215 in it's tracks?
> > > Are you waiting for me to do something/fix something?
> > >
> > > —
> > > You are receiving this because you were mentioned.
> > > Reply to this email directly, view it on GitHub
> > > <#25239 (comment)>,
or
> > mute
> > > the thread
> > > <
> >
>
https://github.com/notifications/unsubscribe-auth/AA95XdPG2FipYrZ16WMRnbpRco3Ptl9Tks5ukrV2gaJpZM4TxrKa
> > >
> > > .
> > >
> > --
> > Giovanni Bajo :: ***@***.***
> > Develer S.r.l. :: http://www.develer.com
> >
> > Tel: +39-055-3984627 (ext: 206)
> >
> > My Blog: http://giovanni.bajo.it
> >
> > —
> > You are receiving this because you commented.
> > Reply to this email directly, view it on GitHub
> > <#25239 (comment)>, or
> mute
> > the thread
> > <
>
https://github.com/notifications/unsubscribe-auth/AeohTkQJvLRRGCytCwhY3it9FVp4k10vks5ukwYogaJpZM4TxrKa
> >
> > .
> >
>
> —
> You are receiving this because you were mentioned.
> Reply to this email directly, view it on GitHub
> <#25239 (comment)>, or
mute
> the thread
> <
https://github.com/notifications/unsubscribe-auth/AA95XQnUJGRSz---5myWIxzfxgxpzlcZks5ukwoOgaJpZM4TxrKa
>
> .
>
--
Giovanni Bajo :: ***@***.***
Develer S.r.l. :: http://www.develer.com
Tel: +39-055-3984627 (ext: 206)
My Blog: http://giovanni.bajo.it
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#25239 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AeohTng5Y5gXrBMwE4dIvVuLZBiKuXeDks5ukw9DgaJpZM4TxrKa>
.
|
|
Thanks for your persistence. And as Giovanni said, don’t hesitate to ping occasionally. All the reviewers are all too human—busy, distractible, forgetful, imperfect. :) |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
We currently emit fix-up code for signed division by -1 as part of genssa.
However, we may be able to prove during the prove pass or generic rules that the divisor cannot be -1. If so, we can save that info and use it during genssa. The Go 1.11 cycle introduced some such optimizations for some math/bits intrinsics and shifts.
This might also provide a tidy means to accomplish this TODO from amd64/ssa.go:
cc @rasky
The text was updated successfully, but these errors were encountered: