| Message ID | 55E5FEE7.1000801@redhat.com |
|---|---|
| State | New |
| Headers | show |
Hi Vladimir, On 1 September 2015 at 21:39, Vladimir Makarov <vmakarov@redhat.com> wrote: > The following patch is for > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 > > The patch was bootstrapped and tested on x86 and x86-64. > > Committed as rev. 227382. > Since this patch, I can see: gcc.dg/vect/slp-perm-5.c (internal compiler error) gcc.dg/vect/slp-perm-5.c -flto -ffat-lto-objects (internal compiler error) on arm* targets. Can you have a look? Thanks, Christophe. > 2015-09-01 Vladimir Makarov <vmakarov@redhat.com> > > PR target/61578 > * lra-lives.c (process_bb_lives): Process move pseudos with the > same value for copies and preferences > * lra-constraints.c (match_reload): Create match reload pseudo > with the same value from single dying input pseudo. >
On 09/02/2015 11:32 AM, Christophe Lyon wrote: > Hi Vladimir, > > > > On 1 September 2015 at 21:39, Vladimir Makarov <vmakarov@redhat.com> wrote: >> The following patch is for >> >> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 >> >> The patch was bootstrapped and tested on x86 and x86-64. >> >> Committed as rev. 227382. >> > Since this patch, I can see: > gcc.dg/vect/slp-perm-5.c (internal compiler error) > gcc.dg/vect/slp-perm-5.c -flto -ffat-lto-objects (internal compiler error) > > on arm* targets. > > Can you have a look? > > Thanks for reporting this, Christophe. Sure, I'll investigate it. It is my fault I should fix it.
On 09/03/2015 11:00 AM, Vladimir Makarov wrote: > On 09/02/2015 11:32 AM, Christophe Lyon wrote: >> Hi Vladimir, >> >> >> >> On 1 September 2015 at 21:39, Vladimir Makarov <vmakarov@redhat.com> >> wrote: >>> The following patch is for >>> >>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 >>> >>> The patch was bootstrapped and tested on x86 and x86-64. >>> >>> Committed as rev. 227382. >>> >> Since this patch, I can see: >> gcc.dg/vect/slp-perm-5.c (internal compiler error) >> gcc.dg/vect/slp-perm-5.c -flto -ffat-lto-objects (internal >> compiler error) >> >> on arm* targets. >> Christophe, I can not reproduce it on my arm board (odroid xu4). Could you provide more info (target, configure options). Thanks.
On 3 September 2015 at 20:28, Vladimir Makarov <vmakarov@redhat.com> wrote: > On 09/03/2015 11:00 AM, Vladimir Makarov wrote: >> >> On 09/02/2015 11:32 AM, Christophe Lyon wrote: >>> >>> Hi Vladimir, >>> >>> >>> >>> On 1 September 2015 at 21:39, Vladimir Makarov <vmakarov@redhat.com> >>> wrote: >>>> >>>> The following patch is for >>>> >>>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 >>>> >>>> The patch was bootstrapped and tested on x86 and x86-64. >>>> >>>> Committed as rev. 227382. >>>> >>> Since this patch, I can see: >>> gcc.dg/vect/slp-perm-5.c (internal compiler error) >>> gcc.dg/vect/slp-perm-5.c -flto -ffat-lto-objects (internal compiler >>> error) >>> >>> on arm* targets. >>> > Christophe, I can not reproduce it on my arm board (odroid xu4). Could you > provide more info (target, configure options). Thanks. > For instance, you can try --target=arm-none-linux-gnueabihf --with-mode=arm --with-cpu=cortex-a9 --with-fpu=neon-fp16
On Tue, Sep 01, 2015 at 03:39:19PM -0400, Vladimir Makarov wrote: > The following patch is for > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 > > The patch was bootstrapped and tested on x86 and x86-64. > > Committed as rev. 227382. > > 2015-09-01 Vladimir Makarov <vmakarov@redhat.com> > > PR target/61578 > * lra-lives.c (process_bb_lives): Process move pseudos with the > same value for copies and preferences > * lra-constraints.c (match_reload): Create match reload pseudo > with the same value from single dying input pseudo. This check-in caused a regression on s390, please see https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 for details. Ciao Dominik ^_^ ^_^
Index: lra-constraints.c =================================================================== --- lra-constraints.c (revision 227381) +++ lra-constraints.c (working copy) @@ -928,10 +928,12 @@ match_reload (signed char out, signed ch they live in the same place. When we create a pseudo we assign value of original pseudo (if any) from which we created the new pseudo. If we create the pseudo from the - input pseudo, the new pseudo will no conflict with the input - pseudo which is wrong when the input pseudo lives after the - insn and as the new pseudo value is changed by the insn - output. Therefore we create the new pseudo from the output. + input pseudo, the new pseudo will have no conflict with the + input pseudo which is wrong when the input pseudo lives after + the insn and as the new pseudo value is changed by the insn + output. Therefore we create the new pseudo from the output + except the case when we have single matched dying input + pseudo. We cannot reuse the current output register because we might have a situation like "a <- a op b", where the constraints @@ -940,8 +942,12 @@ match_reload (signed char out, signed ch so that it doesn't clobber the current value of "a". */ new_in_reg = new_out_reg - = lra_create_new_reg_with_unique_value (outmode, out_rtx, - goal_class, ""); + = (ins[1] < 0 && REG_P (in_rtx) + && (int) REGNO (in_rtx) < lra_new_regno_start + && find_regno_note (curr_insn, REG_DEAD, REGNO (in_rtx)) + ? lra_create_new_reg (inmode, in_rtx, goal_class, "") + : lra_create_new_reg_with_unique_value (outmode, out_rtx, + goal_class, "")); } /* In operand can be got from transformations before processing insn constraints. One example of such transformations is subreg Index: lra-lives.c =================================================================== --- lra-lives.c (revision 227381) +++ lra-lives.c (working copy) @@ -726,28 +726,33 @@ process_bb_lives (basic_block bb, int &c lra_hard_reg_usage[reg->regno] += freq; call_p = CALL_P (curr_insn); + src_regno = (set != NULL_RTX && REG_P (SET_SRC (set)) + ? REGNO (SET_SRC (set)) : -1); + dst_regno = (set != NULL_RTX && REG_P (SET_DEST (set)) + ? REGNO (SET_DEST (set)) : -1); if (complete_info_p - && set != NULL_RTX - && REG_P (SET_DEST (set)) && REG_P (SET_SRC (set)) + && src_regno >= 0 && dst_regno >= 0 /* Check that source regno does not conflict with destination regno to exclude most impossible preferences. */ - && ((((src_regno = REGNO (SET_SRC (set))) >= FIRST_PSEUDO_REGISTER - && ! sparseset_bit_p (pseudos_live, src_regno)) + && (((src_regno >= FIRST_PSEUDO_REGISTER + && (! sparseset_bit_p (pseudos_live, src_regno) + || (dst_regno >= FIRST_PSEUDO_REGISTER + && lra_reg_val_equal_p (src_regno, + lra_reg_info[dst_regno].val, + lra_reg_info[dst_regno].offset)))) || (src_regno < FIRST_PSEUDO_REGISTER && ! TEST_HARD_REG_BIT (hard_regs_live, src_regno))) /* It might be 'inheritance pseudo <- reload pseudo'. */ || (src_regno >= lra_constraint_new_regno_start - && ((int) REGNO (SET_DEST (set)) - >= lra_constraint_new_regno_start) + && dst_regno >= lra_constraint_new_regno_start /* Remember to skip special cases where src/dest regnos are the same, e.g. insn SET pattern has matching constraints like =r,0. */ - && src_regno != (int) REGNO (SET_DEST (set))))) + && src_regno != dst_regno))) { int hard_regno = -1, regno = -1; - dst_regno = REGNO (SET_DEST (set)); if (dst_regno >= lra_constraint_new_regno_start && src_regno >= lra_constraint_new_regno_start) {
The following patch is for https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61578 The patch was bootstrapped and tested on x86 and x86-64. Committed as rev. 227382. 2015-09-01 Vladimir Makarov <vmakarov@redhat.com> PR target/61578 * lra-lives.c (process_bb_lives): Process move pseudos with the same value for copies and preferences * lra-constraints.c (match_reload): Create match reload pseudo with the same value from single dying input pseudo.