IRA copy heuristics

2008-08-31 Thread Richard Sandiford
As I mentioned in: http://gcc.gnu.org/ml/gcc/2008-08/msg00476.html I'd been working on a MIPS IRA port, but got side-tracked by a wrong-code regression. The regression was caused by incorrect EH liveness information. I tried to "fix" it by replacing the current note_stores-based forward scan

Re: IRA copy heuristics

2008-09-02 Thread Vladimir Makarov
Richard Sandiford wrote: > As I mentioned in: > > http://gcc.gnu.org/ml/gcc/2008-08/msg00476.html > > I'd been working on a MIPS IRA port, but got side-tracked by a wrong-code > regression. > > The regression was caused by incorrect EH liveness information. I tried > to "fix" it by replacing

Re: IRA copy heuristics

2008-09-02 Thread Richard Sandiford
Vladimir Makarov <[EMAIL PROTECTED]> writes: > Richard Sandiford wrote: > > > As I mentioned in: > > > > http://gcc.gnu.org/ml/gcc/2008-08/msg00476.html > > > > I'd been working on a MIPS IRA port, but got side-tracked by a wrong-code > > regression. > > > > The regression was caused by

Re: IRA copy heuristics

2008-09-03 Thread H.J. Lu
On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: >> >> If using DF seems like the Right Thing, we could simply apply both >> patches, which would give a similar same allocno order to the one >> we have now. But it seemed better to look a bit deeper first... >> > > Richar

Re: IRA copy heuristics

2008-09-03 Thread Richard Sandiford
"H.J. Lu" <[EMAIL PROTECTED]> writes: > On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: >>> >>> If using DF seems like the Right Thing, we could simply apply both >>> patches, which would give a similar same allocno order to the one >>> we have now. But it seemed better

Re: IRA copy heuristics

2008-09-03 Thread H.J. Lu
On Wed, Sep 3, 2008 at 11:19 AM, Richard Sandiford <[EMAIL PROTECTED]> wrote: > "H.J. Lu" <[EMAIL PROTECTED]> writes: >> On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: If using DF seems like the Right Thing, we could simply apply both patches, which would

Re: IRA copy heuristics

2008-09-03 Thread Vladimir Makarov
Richard Sandiford wrote: "H.J. Lu" <[EMAIL PROTECTED]> writes: On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: If using DF seems like the Right Thing, we could simply apply both patches, which would give a similar same allocno order to the one we have now. B

Re: IRA copy heuristics

2008-09-03 Thread Richard Sandiford
"H.J. Lu" <[EMAIL PROTECTED]> writes: > On Wed, Sep 3, 2008 at 11:19 AM, Richard Sandiford > <[EMAIL PROTECTED]> wrote: >> "H.J. Lu" <[EMAIL PROTECTED]> writes: >>> On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: > > If using DF seems like the Right Thing, we cou

Re: IRA copy heuristics

2008-09-03 Thread Richard Sandiford
Vladimir Makarov <[EMAIL PROTECTED]> writes: > Richard Sandiford wrote: >> "H.J. Lu" <[EMAIL PROTECTED]> writes: >> >>> On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: >>> > If using DF seems like the Right Thing, we could simply apply both > patches, whi

Re: IRA copy heuristics

2008-09-03 Thread Vladimir Makarov
Richard Sandiford wrote: "H.J. Lu" <[EMAIL PROTECTED]> writes: On Wed, Sep 3, 2008 at 11:19 AM, Richard Sandiford <[EMAIL PROTECTED]> wrote: "H.J. Lu" <[EMAIL PROTECTED]> writes: On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: If using DF se

Re: IRA copy heuristics

2008-09-03 Thread Vladimir Makarov
Richard Sandiford wrote: Vladimir Makarov <[EMAIL PROTECTED]> writes: Richard Sandiford wrote: "H.J. Lu" <[EMAIL PROTECTED]> writes: On Tue, Sep 2, 2008 at 8:37 AM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: If using DF seems like the Right Thing, we could sim

Re: IRA copy heuristics

2008-09-03 Thread Richard Sandiford
Vladimir Makarov <[EMAIL PROTECTED]> writes: > Richard Sandiford wrote: >> But as I said to HJ, I'm happy to apply the DF patch in isolation, >> as long as we accept that the benefit of fixing a correctness >> regression outweighs the potential performance regression. >> > Sure, regression is mo

Re: IRA copy heuristics

2008-09-04 Thread Richard Sandiford
Richard Sandiford <[EMAIL PROTECTED]> writes: > Vladimir Makarov <[EMAIL PROTECTED]> writes: >> Richard Sandiford wrote: >>> But as I said to HJ, I'm happy to apply the DF patch in isolation, >>> as long as we accept that the benefit of fixing a correctness >>> regression outweighs the potential pe

Re: IRA copy heuristics

2008-09-04 Thread Vladimir Makarov
Richard Sandiford wrote: Richard Sandiford <[EMAIL PROTECTED]> writes: Vladimir Makarov <[EMAIL PROTECTED]> writes: Richard Sandiford wrote: But as I said to HJ, I'm happy to apply the DF patch in isolation, as long as we accept that the benefit of fixing a correctness regressio

Re: IRA copy heuristics

2008-09-04 Thread David Edelsohn
On Thu, Sep 4, 2008 at 7:39 PM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: > Meanwhile I am going to submit your second patch with an added > comment. The patch permits gcc to generate the same quality code as > before your first patch. Why? As Richard said before: "... it changes the heuristi

Re: IRA copy heuristics

2008-09-04 Thread Peter Bergner
On Thu, 2008-09-04 at 20:28 -0400, David Edelsohn wrote: > On Thu, Sep 4, 2008 at 7:39 PM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: > > Meanwhile I am going to submit your second patch with an added > > comment. The patch permits gcc to generate the same quality code as > > before your first pa

Re: IRA copy heuristics

2008-09-04 Thread Jeff Law
Peter Bergner wrote: On Thu, 2008-09-04 at 20:28 -0400, David Edelsohn wrote: On Thu, Sep 4, 2008 at 7:39 PM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: Meanwhile I am going to submit your second patch with an added comment. The patch permits gcc to generate the same quality code as b

Re: IRA copy heuristics

2008-09-05 Thread Vladimir Makarov
David Edelsohn wrote: On Thu, Sep 4, 2008 at 7:39 PM, Vladimir Makarov <[EMAIL PROTECTED]> wrote: Meanwhile I am going to submit your second patch with an added comment. The patch permits gcc to generate the same quality code as before your first patch. Why? As Richard said before: