Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-02-02 Thread Jakub Jelinek
On Thu, Feb 02, 2017 at 02:47:12AM -0600, Segher Boessenkool wrote: > On Thu, Feb 02, 2017 at 08:51:19AM +0100, Jakub Jelinek wrote: > > Thanks, though at this point I think the > > http://gcc.gnu.org/ml/gcc-patches/2017-01/msg02390.html > > patch is better. With the earliest patch (the one below)

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-02-02 Thread Segher Boessenkool
On Thu, Feb 02, 2017 at 08:51:19AM +0100, Jakub Jelinek wrote: > Thanks, though at this point I think the > http://gcc.gnu.org/ml/gcc-patches/2017-01/msg02390.html > patch is better. With the earliest patch (the one below) for -mvsx > -mno-popcntd (or whatever -mcpu= implies that) there is no DF->

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-02-01 Thread Jakub Jelinek
On Wed, Feb 01, 2017 at 11:56:34PM -0600, Segher Boessenkool wrote: > On Mon, Jan 30, 2017 at 10:27:15PM +0100, Jakub Jelinek wrote: > > Accoring to make mddump generated tmp-mddump.md, on powerpc the only pattern > > with unsigned_fix:DI where the inner operand is SF or DFmode is the > > *fixuns_t

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-02-01 Thread Segher Boessenkool
On Tue, Jan 31, 2017 at 09:11:11AM +0100, Jakub Jelinek wrote: > What I think this patch doesn't handle properly is SFmode -> DImode > unsigned_fix. I have no idea how SFmode is represented in vector registers, > so have no idea if xscvdpuxds actually handles SFmode as well. All floating point sc

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-02-01 Thread Segher Boessenkool
On Mon, Jan 30, 2017 at 10:27:15PM +0100, Jakub Jelinek wrote: > Accoring to make mddump generated tmp-mddump.md, on powerpc the only pattern > with unsigned_fix:DI where the inner operand is SF or DFmode is the > *fixuns_truncdi2_fctiduz. There is an expander for that instruction, > which uses di

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-01-31 Thread Jakub Jelinek
On Tue, Jan 31, 2017 at 09:11:11AM +0100, Jakub Jelinek wrote: > But I don't see an iterator that would handle SFmode in vsx.md, only > VSX_B is only > (define_mode_iterator VSX_B [DF V4SF V2DF]) > and even most of the iterators the vsx.md insn uses don't handle SFmode. > So, shall the condition ac

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-01-31 Thread Jakub Jelinek
On Mon, Jan 30, 2017 at 06:43:09PM -0600, Segher Boessenkool wrote: > Hi Jakub, Mike, > > On Mon, Jan 30, 2017 at 10:27:15PM +0100, Jakub Jelinek wrote: > > Accoring to make mddump generated tmp-mddump.md, on powerpc the only pattern > > with unsigned_fix:DI where the inner operand is SF or DFmode

Re: [PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-01-30 Thread Segher Boessenkool
Hi Jakub, Mike, On Mon, Jan 30, 2017 at 10:27:15PM +0100, Jakub Jelinek wrote: > Accoring to make mddump generated tmp-mddump.md, on powerpc the only pattern > with unsigned_fix:DI where the inner operand is SF or DFmode is the > *fixuns_truncdi2_fctiduz. It seems like vsx_fixuns_trunc2 (in vsx.m

[PATCH] Fix ICEs with power8 fixuns_truncdi2 patterns (PR target/79197)

2017-01-30 Thread Jakub Jelinek
Hi! Accoring to make mddump generated tmp-mddump.md, on powerpc the only pattern with unsigned_fix:DI where the inner operand is SF or DFmode is the *fixuns_truncdi2_fctiduz. There is an expander for that instruction, which uses different operand predicates and different condition, so the followi