| Message ID | 7FB04A5C213E9943A72EE127DB74F0ADA6899756DA@SJEXCHCCR02.corp.ad.broadcom.com |
|---|---|
| State | New |
| Headers | show |
On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> wrote: > 2010-08-04 Bingfeng Mei <bmei@broadcom.com> > > * alias.c (walk_mems_2): Call rtx_refs_may_alias_p > instead of alias_sets_conflict_p to get more accurate > alias set information. OK. Diego.
On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo <dnovillo@google.com> wrote: > On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> wrote: > >> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> >> >> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p >> instead of alias_sets_conflict_p to get more accurate >> alias set information. > > OK. Wait. You need to use a proper dependence function instead, rtx_refs_may_alias_p isn't supposed to be used directly. Richard. > > Diego. >
On Wed, Aug 4, 2010 at 4:57 PM, Richard Guenther <richard.guenther@gmail.com> wrote: > On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo <dnovillo@google.com> wrote: >> On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> wrote: >> >>> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> >>> >>> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p >>> instead of alias_sets_conflict_p to get more accurate >>> alias set information. >> >> OK. > > Wait. You need to use a proper dependence function instead, > rtx_refs_may_alias_p isn't supposed to be used directly. Which would be true_dependence if using TBAA is valid here (which I am not sure). Richard. > Richard. > >> >> Diego. >> >
On Wed, Aug 4, 2010 at 4:58 PM, Richard Guenther <richard.guenther@gmail.com> wrote: > On Wed, Aug 4, 2010 at 4:57 PM, Richard Guenther > <richard.guenther@gmail.com> wrote: >> On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo <dnovillo@google.com> wrote: >>> On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> wrote: >>> >>>> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> >>>> >>>> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p >>>> instead of alias_sets_conflict_p to get more accurate >>>> alias set information. >>> >>> OK. >> >> Wait. You need to use a proper dependence function instead, >> rtx_refs_may_alias_p isn't supposed to be used directly. > > Which would be true_dependence if using TBAA is valid here > (which I am not sure). After this patch insn_alias_sets_conflict_p is also seriously misnamed. As it is used only from within ddg.c it should be moved there and made private. /* Given two nodes, analyze their RTL insns and add inter-loop mem deps to ddg G. */ static void add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) { if (!insn_alias_sets_conflict_p (from->insn, to->insn)) /* Do not create edge if memory references have disjoint alias sets. */ return; the comment ("inter-loop mem deps") suggests that using TBAA is not valid here. Richard. > Richard. > >> Richard. >> >>> >>> Diego. >>> >> >
> -----Original Message----- > From: Richard Guenther [mailto:richard.guenther@gmail.com] > Sent: 04 August 2010 16:02 > To: Diego Novillo > Cc: Bingfeng Mei; gcc-patches@gcc.gnu.org > Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > alias_sets_conflict_p in > > On Wed, Aug 4, 2010 at 4:58 PM, Richard Guenther > <richard.guenther@gmail.com> wrote: > > On Wed, Aug 4, 2010 at 4:57 PM, Richard Guenther > > <richard.guenther@gmail.com> wrote: > >> On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo <dnovillo@google.com> > wrote: > >>> On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> > wrote: > >>> > >>>> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> > >>>> > >>>> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p > >>>> instead of alias_sets_conflict_p to get more accurate > >>>> alias set information. > >>> > >>> OK. > >> > >> Wait. You need to use a proper dependence function instead, > >> rtx_refs_may_alias_p isn't supposed to be used directly. > > > > Which would be true_dependence if using TBAA is valid here > > (which I am not sure). > > After this patch insn_alias_sets_conflict_p is also seriously > misnamed. As it is used only from within ddg.c it should > be moved there and made private. So I need to make rtx_refs_may_alias_p global? > > /* Given two nodes, analyze their RTL insns and add inter-loop mem deps > to ddg G. */ > static void > add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) > { > if (!insn_alias_sets_conflict_p (from->insn, to->insn)) > /* Do not create edge if memory references have disjoint alias sets. > */ > return; > > the comment ("inter-loop mem deps") suggests that using TBAA > is not valid here. Does TBAA mean type-based alias analysis? Why is it not valid here? > > Richard. > > > Richard. > > > >> Richard. > >> > >>> > >>> Diego. > >>> > >> > >
On Wed, Aug 4, 2010 at 5:06 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > > >> -----Original Message----- >> From: Richard Guenther [mailto:richard.guenther@gmail.com] >> Sent: 04 August 2010 16:02 >> To: Diego Novillo >> Cc: Bingfeng Mei; gcc-patches@gcc.gnu.org >> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of >> alias_sets_conflict_p in >> >> On Wed, Aug 4, 2010 at 4:58 PM, Richard Guenther >> <richard.guenther@gmail.com> wrote: >> > On Wed, Aug 4, 2010 at 4:57 PM, Richard Guenther >> > <richard.guenther@gmail.com> wrote: >> >> On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo <dnovillo@google.com> >> wrote: >> >>> On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> >> wrote: >> >>> >> >>>> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> >> >>>> >> >>>> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p >> >>>> instead of alias_sets_conflict_p to get more accurate >> >>>> alias set information. >> >>> >> >>> OK. >> >> >> >> Wait. You need to use a proper dependence function instead, >> >> rtx_refs_may_alias_p isn't supposed to be used directly. >> > >> > Which would be true_dependence if using TBAA is valid here >> > (which I am not sure). >> >> After this patch insn_alias_sets_conflict_p is also seriously >> misnamed. As it is used only from within ddg.c it should >> be moved there and made private. > > So I need to make rtx_refs_may_alias_p global? No. You need to adjust add_inter_loop_mem_dep to, for the cases of {read,write} vs. {read,write} use the proper dependence queires Like /* Given two nodes, analyze their RTL insns and add inter-loop mem deps to ddg G. */ static void add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) { - if (!insn_alias_sets_conflict_p (from->insn, to->insn)) - /* Do not create edge if memory references have disjoint alias sets. */ - return; if (mem_write_insn_p (from->insn)) { if (mem_read_insn_p (to->insn)) + { + if (XXX_dependence (the-mems-in from->insn, the-mems-in to->insn)) create_ddg_dep_no_link (g, from, to, DEBUG_INSN_P (to->insn) ? ANTI_DEP : TRUE_DEP, MEM_DEP, 1); ... with XXX being output or anti or true. etc. Richard. >> >> /* Given two nodes, analyze their RTL insns and add inter-loop mem deps >> to ddg G. */ >> static void >> add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) >> { >> if (!insn_alias_sets_conflict_p (from->insn, to->insn)) >> /* Do not create edge if memory references have disjoint alias sets. >> */ >> return; >> >> the comment ("inter-loop mem deps") suggests that using TBAA >> is not valid here. > Does TBAA mean type-based alias analysis? Why is it not valid here? > >> >> Richard. >> >> > Richard. >> > >> >> Richard. >> >> >> >>> >> >>> Diego. >> >>> >> >> >> > > > >
> -----Original Message----- > From: Richard Guenther [mailto:richard.guenther@gmail.com] > Sent: 04 August 2010 16:15 > To: Bingfeng Mei > Cc: Diego Novillo; gcc-patches@gcc.gnu.org > Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > alias_sets_conflict_p in > > On Wed, Aug 4, 2010 at 5:06 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > > > > > >> -----Original Message----- > >> From: Richard Guenther [mailto:richard.guenther@gmail.com] > >> Sent: 04 August 2010 16:02 > >> To: Diego Novillo > >> Cc: Bingfeng Mei; gcc-patches@gcc.gnu.org > >> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > >> alias_sets_conflict_p in > >> > >> On Wed, Aug 4, 2010 at 4:58 PM, Richard Guenther > >> <richard.guenther@gmail.com> wrote: > >> > On Wed, Aug 4, 2010 at 4:57 PM, Richard Guenther > >> > <richard.guenther@gmail.com> wrote: > >> >> On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo > <dnovillo@google.com> > >> wrote: > >> >>> On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> > >> wrote: > >> >>> > >> >>>> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> > >> >>>> > >> >>>> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p > >> >>>> instead of alias_sets_conflict_p to get more accurate > >> >>>> alias set information. > >> >>> > >> >>> OK. > >> >> > >> >> Wait. You need to use a proper dependence function instead, > >> >> rtx_refs_may_alias_p isn't supposed to be used directly. > >> > > >> > Which would be true_dependence if using TBAA is valid here > >> > (which I am not sure). > >> > >> After this patch insn_alias_sets_conflict_p is also seriously > >> misnamed. As it is used only from within ddg.c it should > >> be moved there and made private. > > > > So I need to make rtx_refs_may_alias_p global? > > No. You need to adjust add_inter_loop_mem_dep to, for the > cases of {read,write} vs. {read,write} use the proper dependence > queires > > Like > > /* Given two nodes, analyze their RTL insns and add inter-loop mem deps > to ddg G. */ > static void > add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) > { > - if (!insn_alias_sets_conflict_p (from->insn, to->insn)) > - /* Do not create edge if memory references have disjoint alias > sets. */ > - return; > > if (mem_write_insn_p (from->insn)) > { > if (mem_read_insn_p (to->insn)) > + { > + if (XXX_dependence (the-mems-in from->insn, the-mems-in to->insn)) > create_ddg_dep_no_link (g, from, to, > DEBUG_INSN_P (to->insn) > ? ANTI_DEP : TRUE_DEP, MEM_DEP, 1); > ... with XXX being output or anti or true. > > etc. But this is cross-iteration dependence, meaning backward. I assume true_dependece function won't work here. > > Richard. > > >> > >> /* Given two nodes, analyze their RTL insns and add inter-loop mem > deps > >> to ddg G. */ > >> static void > >> add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr > to) > >> { > >> if (!insn_alias_sets_conflict_p (from->insn, to->insn)) > >> /* Do not create edge if memory references have disjoint alias > sets. > >> */ > >> return; > >> > >> the comment ("inter-loop mem deps") suggests that using TBAA > >> is not valid here. > > Does TBAA mean type-based alias analysis? Why is it not valid here? > > > >> > >> Richard. > >> > >> > Richard. > >> > > >> >> Richard. > >> >> > >> >>> > >> >>> Diego. > >> >>> > >> >> > >> > > > > > > >
On Wed, Aug 4, 2010 at 4:52 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > Index: alias.c > =================================================================== > --- alias.c (revision 162821) > +++ alias.c (working copy) > @@ -454,7 +454,7 @@ > { > if (MEM_P (*x)) > { > - if (alias_sets_conflict_p (MEM_ALIAS_SET(*x), MEM_ALIAS_SET(mem))) > + if (rtx_refs_may_alias_p (*x, mem, true)) > return 1; > > return -1; > Hi, Could you create your patches with the diff -p option? That makes it easier to see what function you are patching. Ciao! Steven
Hi, On Wed, 4 Aug 2010, Richard Guenther wrote: > /* Given two nodes, analyze their RTL insns and add inter-loop mem deps > to ddg G. */ > static void > add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) > { > if (!insn_alias_sets_conflict_p (from->insn, to->insn)) > /* Do not create edge if memory references have disjoint alias sets. */ > return; > > the comment ("inter-loop mem deps") suggests that using TBAA > is not valid here. Careful. type-based alias analysis can be used. It's offset-based disambiguation that cannot for cross-iteration references. Unfortunately I don't think the current interface we have make it possible to use TBAA but not also offset-based disambiguation. Ciao, Michael.
Does rtx_refs_may_alias_p include offset based disambiguation (with/without tbaa flag). If yes, this patch is not correct, I need to work out a better one. Cheers, Bingfeng > -----Original Message----- > From: Michael Matz [mailto:matz@suse.de] > Sent: 04 August 2010 16:36 > To: Richard Guenther > Cc: Diego Novillo; Bingfeng Mei; gcc-patches@gcc.gnu.org > Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > alias_sets_conflict_p in > > Hi, > > On Wed, 4 Aug 2010, Richard Guenther wrote: > > > /* Given two nodes, analyze their RTL insns and add inter-loop mem > deps > > to ddg G. */ > > static void > > add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) > > { > > if (!insn_alias_sets_conflict_p (from->insn, to->insn)) > > /* Do not create edge if memory references have disjoint alias > sets. */ > > return; > > > > the comment ("inter-loop mem deps") suggests that using TBAA > > is not valid here. > > Careful. type-based alias analysis can be used. It's offset-based > disambiguation that cannot for cross-iteration references. > > Unfortunately I don't think the current interface we have make it > possible > to use TBAA but not also offset-based disambiguation. > > > Ciao, > Michael.
Hi, On Wed, 4 Aug 2010, Bingfeng Mei wrote: > Does rtx_refs_may_alias_p include offset based disambiguation (with/without > tbaa flag). Actually it doesn't (we switch off recognizing SSA names as being equal when called from RTL land). I forgot that this was recently changed. Unfortunately I was wrong with my assertion that TBAA can be used cross iteration. The catch is the new memory model, with it's asymmetric dependencies: you may move a load before a non-conflicting (type wise) store, but you may not move a store before a non-conflicting load. In cross-iteration situations you are posed with both cases, hence you can't really use type-based disambiguation (perhaps in some few special cases one might, but not generally). Basically for cross iteration analysis you can only use points-to information (or generally everything that is loop invariant for the loop in question, but that isn't implemented). > If yes, this patch is not correct, I need to work out a better one. It would be better to use the four different predicates we have in RTL land as Richard suggested, they do the right thing for each dependency type (switching off TBAA when appropriate). Ciao, Michael.
A general comment. A set of unified high level aliasing/memory disambiguation/data dependence query interfaces can be handy for many different alias information clients such as Scalar optimizer, Loop Transformations, AutoPar, Scheduler, ModSched etc. 1) Instead of returning a bool, it is useful to return aliasing kind: no_alias, may_alias, must_alias, etc. 2) The aliasing query result is in many cases ambiguous without a context, e.g, an intra-iteration no alias could become inter-iteration may-alias. To solve this, the interface need to provide a way for the client code to specify what it wants -- i.e. a bool flag indicating whether it cares about cross iteration/carried dependence or not -- or better yet, passing a loop context -- if the loop context is null, it only cares about intra iteration aliasing. The implementation will need to check the context and decide the most precise answer depending on the source of the alias information. For instance, points-to info is 'context insensitive', offset based tbaa is not either in cases; restrict qualified locals can be tricky --- as depending on the loop in the loop nest that is queried, it may return different answer. 3) Client code (e.g. ModSched) may also care about dependence direction/distance -- the information can be returned if known When it comes to cross iteration aliasing/dependence information (and distance), the information needs to be updated properly with loop unrolling etc. Other useful things to have: 1. debug trace on alias queries -- indicating what alias result is returned and what the source is (points to, tbaa, ...) 2. dbgcnt support for alias query for triaging alias problems. Thanks, David On Wed, Aug 4, 2010 at 8:15 AM, Richard Guenther <richard.guenther@gmail.com> wrote: > On Wed, Aug 4, 2010 at 5:06 PM, Bingfeng Mei <bmei@broadcom.com> wrote: >> >> >>> -----Original Message----- >>> From: Richard Guenther [mailto:richard.guenther@gmail.com] >>> Sent: 04 August 2010 16:02 >>> To: Diego Novillo >>> Cc: Bingfeng Mei; gcc-patches@gcc.gnu.org >>> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of >>> alias_sets_conflict_p in >>> >>> On Wed, Aug 4, 2010 at 4:58 PM, Richard Guenther >>> <richard.guenther@gmail.com> wrote: >>> > On Wed, Aug 4, 2010 at 4:57 PM, Richard Guenther >>> > <richard.guenther@gmail.com> wrote: >>> >> On Wed, Aug 4, 2010 at 4:56 PM, Diego Novillo <dnovillo@google.com> >>> wrote: >>> >>> On Wed, Aug 4, 2010 at 14:52, Bingfeng Mei <bmei@broadcom.com> >>> wrote: >>> >>> >>> >>>> 2010-08-04 Bingfeng Mei <bmei@broadcom.com> >>> >>>> >>> >>>> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p >>> >>>> instead of alias_sets_conflict_p to get more accurate >>> >>>> alias set information. >>> >>> >>> >>> OK. >>> >> >>> >> Wait. You need to use a proper dependence function instead, >>> >> rtx_refs_may_alias_p isn't supposed to be used directly. >>> > >>> > Which would be true_dependence if using TBAA is valid here >>> > (which I am not sure). >>> >>> After this patch insn_alias_sets_conflict_p is also seriously >>> misnamed. As it is used only from within ddg.c it should >>> be moved there and made private. >> >> So I need to make rtx_refs_may_alias_p global? > > No. You need to adjust add_inter_loop_mem_dep to, for the > cases of {read,write} vs. {read,write} use the proper dependence queires > > Like > > /* Given two nodes, analyze their RTL insns and add inter-loop mem deps > to ddg G. */ > static void > add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) > { > - if (!insn_alias_sets_conflict_p (from->insn, to->insn)) > - /* Do not create edge if memory references have disjoint alias sets. */ > - return; > > if (mem_write_insn_p (from->insn)) > { > if (mem_read_insn_p (to->insn)) > + { > + if (XXX_dependence (the-mems-in from->insn, the-mems-in to->insn)) > create_ddg_dep_no_link (g, from, to, > DEBUG_INSN_P (to->insn) > ? ANTI_DEP : TRUE_DEP, MEM_DEP, 1); > ... with XXX being output or anti or true. > > etc. > > Richard. > >>> >>> /* Given two nodes, analyze their RTL insns and add inter-loop mem deps >>> to ddg G. */ >>> static void >>> add_inter_loop_mem_dep (ddg_ptr g, ddg_node_ptr from, ddg_node_ptr to) >>> { >>> if (!insn_alias_sets_conflict_p (from->insn, to->insn)) >>> /* Do not create edge if memory references have disjoint alias sets. >>> */ >>> return; >>> >>> the comment ("inter-loop mem deps") suggests that using TBAA >>> is not valid here. >> Does TBAA mean type-based alias analysis? Why is it not valid here? >> >>> >>> Richard. >>> >>> > Richard. >>> > >>> >> Richard. >>> >> >>> >>> >>> >>> Diego. >>> >>> >>> >> >>> > >> >> >> >
Hi, If we are going to use true_dependence function etc in ddg.c, we should be able to disable loop variant stuff (currently I can only think of offset-based memory disambiguation and TBAA). I added a flag loop_invariant to relevant functions in attached patch. The patch is tested and bootstrapped. Is this OK? Next, I need to adapt ddg.c to use these XXX_dependence functions, move and rename insn_alias_sets_conflict_p & friends to ddg.c. The original add_inter_loop_mem_dep is over-simplified. For example, it doesn't consider parallel patterns with more than one memory expressions. Thanks Bingfeng > -----Original Message----- > From: Michael Matz [mailto:matz@suse.de] > Sent: 04 August 2010 17:15 > To: Bingfeng Mei > Cc: Richard Guenther; Diego Novillo; gcc-patches@gcc.gnu.org > Subject: RE: [PATCH] Use rtx_refs_may_alias_p instead of > alias_sets_conflict_p in > > Hi, > > On Wed, 4 Aug 2010, Bingfeng Mei wrote: > > > Does rtx_refs_may_alias_p include offset based disambiguation > (with/without > > tbaa flag). > > Actually it doesn't (we switch off recognizing SSA names as being equal > when called from RTL land). I forgot that this was recently changed. > > Unfortunately I was wrong with my assertion that TBAA can be used cross > iteration. The catch is the new memory model, with it's asymmetric > dependencies: you may move a load before a non-conflicting (type > wise) store, but you may not move a store before a non-conflicting load. > > In cross-iteration situations you are posed with both cases, hence you > can't really use type-based disambiguation (perhaps in some few special > cases one might, but not generally). > > Basically for cross iteration analysis you can only use points-to > information (or generally everything that is loop invariant for the > loop > in question, but that isn't implemented). > > > If yes, this patch is not correct, I need to work out a better one. > > It would be better to use the four different predicates we have in RTL > land as Richard suggested, they do the right thing for each dependency > type (switching off TBAA when appropriate). > > > Ciao, > Michael.
On Thu, Aug 5, 2010 at 7:00 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > Hi, > If we are going to use true_dependence function etc in ddg.c, > we should be able to disable loop variant stuff (currently > I can only think of offset-based memory disambiguation and TBAA). > I added a flag loop_invariant to relevant functions in attached > patch. The patch is tested and bootstrapped. Is this OK? > > Next, I need to adapt ddg.c to use these XXX_dependence functions, > move and rename insn_alias_sets_conflict_p & friends to ddg.c. > The original add_inter_loop_mem_dep is over-simplified. > For example, it doesn't consider parallel patterns with more than > one memory expressions. + if (!loop_invariant && &&s go to the next line. if (DIFFERENT_ALIAS_SETS_P (x, mem)) return 0; - if (nonoverlapping_memrefs_p (mem, x)) + if (nonoverlapping_memrefs_p (mem, x, loop_invariant)) return 0; The DIFFERENT_ALIAS_SETS_P would also need guarding. I think that instead of changing all present functions it would be better to create a new entry to the alias oracle for inter-iteration queries, as for example you can't distinguish true and anti dependence (which is also why TBAA isn't valid here). In fact the implementation would be the same for all of RW, WR and WW dependence types. The query would exactly match what Zdenek thinks of a may-alias query, so I'd suggest to simply name it may_alias_p. Richard. > Thanks > Bingfeng > >> -----Original Message----- >> From: Michael Matz [mailto:matz@suse.de] >> Sent: 04 August 2010 17:15 >> To: Bingfeng Mei >> Cc: Richard Guenther; Diego Novillo; gcc-patches@gcc.gnu.org >> Subject: RE: [PATCH] Use rtx_refs_may_alias_p instead of >> alias_sets_conflict_p in >> >> Hi, >> >> On Wed, 4 Aug 2010, Bingfeng Mei wrote: >> >> > Does rtx_refs_may_alias_p include offset based disambiguation >> (with/without >> > tbaa flag). >> >> Actually it doesn't (we switch off recognizing SSA names as being equal >> when called from RTL land). I forgot that this was recently changed. >> >> Unfortunately I was wrong with my assertion that TBAA can be used cross >> iteration. The catch is the new memory model, with it's asymmetric >> dependencies: you may move a load before a non-conflicting (type >> wise) store, but you may not move a store before a non-conflicting load. >> >> In cross-iteration situations you are posed with both cases, hence you >> can't really use type-based disambiguation (perhaps in some few special >> cases one might, but not generally). >> >> Basically for cross iteration analysis you can only use points-to >> information (or generally everything that is loop invariant for the >> loop >> in question, but that isn't implemented). >> >> > If yes, this patch is not correct, I need to work out a better one. >> >> It would be better to use the four different predicates we have in RTL >> land as Richard suggested, they do the right thing for each dependency >> type (switching off TBAA when appropriate). >> >> >> Ciao, >> Michael. > >
Hi,
I created a new may_alias_p function as Richard proposed. The function
disables offset-based memory disambiguation and TBAA, it is also slightly
more conservative than true/output/anti_dependence. For example,
write_dependence checks:
if (GET_MODE (x) == BLKmode && GET_CODE (XEXP (x, 0)) == SCRATCH)
return 1;
if (GET_MODE (mem) == BLKmode && GET_CODE (XEXP (mem, 0)) == SCRATCH)
return 1;
true_depdence checks both above and following (isn't above
statement redundant then?).
if (mem_mode == BLKmode || GET_MODE (x) == BLKmode)
return 1;
may_alias_p takes more conservative check of the latter one only.
Since insn_alias_sets_conflict_p is misnamed and only used once by ddg.c,
it & friends (walk_mems_1, walk_mems_2) are moved into ddg.c and
renamed as insns_may_alias_p.
The patch is bootstrapped on x86_64-unknown-linux-gnu and tested.
OK for trunk?
Thanks,
Bingfeng
2010-08-10 Bingfeng Mei <bmei@broadcom.com>
* ddg.c (walk_mems_2): Moved from alias.c, use may_alias_p instead of
alias_sets_conflict_p.
(walk_mems_1): Moved from alias.c.
(insns_may_alias_p): New function, originally insn_alias_sets_conflict_p
in alias.c.
(add_inter_loop_mem_dep): Use insns_may_alias_p now.
* cse.c (cse_insn): New argument in calling nonoverlapping_memrefs_p.
* alias.c (walk_mems_2): Moved to ddg.c.
(walk_mems_1): Ditto.
(insn_alias_sets_conflict_p): Renamed to insns_may_alias_p and moved
to ddg.c.
(nonoverlapping_memrefs_p): Add flag to guard offset-based memory
disambiguation.
*(may_alias_p): New function to check whether two memory expression
may alias or not. Currently used in buidling inter-iteration memory
dependence.
*alias.h (nonoverlapping_memrefs_p): New flag as third argument.
(insn_alias_sets_conflict_p): Removed
*rtl.h (may_alias_p): New function prototype.
> -----Original Message-----
> From: Richard Guenther [mailto:richard.guenther@gmail.com]
> Sent: 06 August 2010 11:10
> To: Bingfeng Mei
> Cc: Michael Matz; Diego Novillo; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of
> alias_sets_conflict_p in
>
> On Thu, Aug 5, 2010 at 7:00 PM, Bingfeng Mei <bmei@broadcom.com> wrote:
> > Hi,
> > If we are going to use true_dependence function etc in ddg.c,
> > we should be able to disable loop variant stuff (currently
> > I can only think of offset-based memory disambiguation and TBAA).
> > I added a flag loop_invariant to relevant functions in attached
> > patch. The patch is tested and bootstrapped. Is this OK?
> >
> > Next, I need to adapt ddg.c to use these XXX_dependence functions,
> > move and rename insn_alias_sets_conflict_p & friends to ddg.c.
> > The original add_inter_loop_mem_dep is over-simplified.
> > For example, it doesn't consider parallel patterns with more than
> > one memory expressions.
>
> + if (!loop_invariant &&
>
> &&s go to the next line.
>
> if (DIFFERENT_ALIAS_SETS_P (x, mem))
> return 0;
>
> - if (nonoverlapping_memrefs_p (mem, x))
> + if (nonoverlapping_memrefs_p (mem, x, loop_invariant))
> return 0;
>
> The DIFFERENT_ALIAS_SETS_P would also need guarding.
>
> I think that instead of changing all present functions it would be
> better to create a new entry to the alias oracle for inter-iteration
> queries, as for example you can't distinguish true and anti dependence
> (which is also why TBAA isn't valid here). In fact the implementation
> would be the same for all of RW, WR and WW dependence types.
>
> The query would exactly match what Zdenek thinks of a
> may-alias query, so I'd suggest to simply name it may_alias_p.
>
> Richard.
>
> > Thanks
> > Bingfeng
> >
> >> -----Original Message-----
> >> From: Michael Matz [mailto:matz@suse.de]
> >> Sent: 04 August 2010 17:15
> >> To: Bingfeng Mei
> >> Cc: Richard Guenther; Diego Novillo; gcc-patches@gcc.gnu.org
> >> Subject: RE: [PATCH] Use rtx_refs_may_alias_p instead of
> >> alias_sets_conflict_p in
> >>
> >> Hi,
> >>
> >> On Wed, 4 Aug 2010, Bingfeng Mei wrote:
> >>
> >> > Does rtx_refs_may_alias_p include offset based disambiguation
> >> (with/without
> >> > tbaa flag).
> >>
> >> Actually it doesn't (we switch off recognizing SSA names as being
> equal
> >> when called from RTL land). I forgot that this was recently changed.
> >>
> >> Unfortunately I was wrong with my assertion that TBAA can be used
> cross
> >> iteration. The catch is the new memory model, with it's asymmetric
> >> dependencies: you may move a load before a non-conflicting (type
> >> wise) store, but you may not move a store before a non-conflicting
> load.
> >>
> >> In cross-iteration situations you are posed with both cases, hence
> you
> >> can't really use type-based disambiguation (perhaps in some few
> special
> >> cases one might, but not generally).
> >>
> >> Basically for cross iteration analysis you can only use points-to
> >> information (or generally everything that is loop invariant for the
> >> loop
> >> in question, but that isn't implemented).
> >>
> >> > If yes, this patch is not correct, I need to work out a better one.
> >>
> >> It would be better to use the four different predicates we have in
> RTL
> >> land as Richard suggested, they do the right thing for each
> dependency
> >> type (switching off TBAA when appropriate).
> >>
> >>
> >> Ciao,
> >> Michael.
> >
> >
On Mon, Aug 9, 2010 at 2:41 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > Hi, > I created a new may_alias_p function as Richard proposed. The function > disables offset-based memory disambiguation and TBAA, it is also slightly > more conservative than true/output/anti_dependence. For example, > write_dependence checks: > if (GET_MODE (x) == BLKmode && GET_CODE (XEXP (x, 0)) == SCRATCH) > return 1; > if (GET_MODE (mem) == BLKmode && GET_CODE (XEXP (mem, 0)) == SCRATCH) > return 1; > > true_depdence checks both above and following (isn't above > statement redundant then?). > if (mem_mode == BLKmode || GET_MODE (x) == BLKmode) > return 1; > may_alias_p takes more conservative check of the latter one only. > > Since insn_alias_sets_conflict_p is misnamed and only used once by ddg.c, > it & friends (walk_mems_1, walk_mems_2) are moved into ddg.c and > renamed as insns_may_alias_p. > > The patch is bootstrapped on x86_64-unknown-linux-gnu and tested. > OK for trunk? @@ -2190,7 +2153,7 @@ adjust_offset_for_component_ref (tree x, X and Y and they do not overlap. */ int -nonoverlapping_memrefs_p (const_rtx x, const_rtx y) +nonoverlapping_memrefs_p (const_rtx x, const_rtx y, bool loop_invariant) the new parameter needs documenting. + +/* Check whether x may be aliased with mem. Don't do offset-based + memory disambiguation & TBAA*/ +int +may_alias_p (const_rtx mem, const_rtx x) two spaces after '.', a full-stop and two spaces at the end of the comment. 'x' and 'mem' should be all-caps. Otherwise the patch looks fine to me. Thus, the patch is ok with the above changes. Thanks, Richard. > Thanks, > Bingfeng > > > 2010-08-10 Bingfeng Mei <bmei@broadcom.com> > > * ddg.c (walk_mems_2): Moved from alias.c, use may_alias_p instead of > alias_sets_conflict_p. > (walk_mems_1): Moved from alias.c. > (insns_may_alias_p): New function, originally insn_alias_sets_conflict_p > in alias.c. > (add_inter_loop_mem_dep): Use insns_may_alias_p now. > * cse.c (cse_insn): New argument in calling nonoverlapping_memrefs_p. > * alias.c (walk_mems_2): Moved to ddg.c. > (walk_mems_1): Ditto. > (insn_alias_sets_conflict_p): Renamed to insns_may_alias_p and moved > to ddg.c. > (nonoverlapping_memrefs_p): Add flag to guard offset-based memory > disambiguation. > *(may_alias_p): New function to check whether two memory expression > may alias or not. Currently used in buidling inter-iteration memory > dependence. > *alias.h (nonoverlapping_memrefs_p): New flag as third argument. > (insn_alias_sets_conflict_p): Removed > *rtl.h (may_alias_p): New function prototype. > >> -----Original Message----- >> From: Richard Guenther [mailto:richard.guenther@gmail.com] >> Sent: 06 August 2010 11:10 >> To: Bingfeng Mei >> Cc: Michael Matz; Diego Novillo; gcc-patches@gcc.gnu.org >> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of >> alias_sets_conflict_p in >> >> On Thu, Aug 5, 2010 at 7:00 PM, Bingfeng Mei <bmei@broadcom.com> wrote: >> > Hi, >> > If we are going to use true_dependence function etc in ddg.c, >> > we should be able to disable loop variant stuff (currently >> > I can only think of offset-based memory disambiguation and TBAA). >> > I added a flag loop_invariant to relevant functions in attached >> > patch. The patch is tested and bootstrapped. Is this OK? >> > >> > Next, I need to adapt ddg.c to use these XXX_dependence functions, >> > move and rename insn_alias_sets_conflict_p & friends to ddg.c. >> > The original add_inter_loop_mem_dep is over-simplified. >> > For example, it doesn't consider parallel patterns with more than >> > one memory expressions. >> >> + if (!loop_invariant && >> >> &&s go to the next line. >> >> if (DIFFERENT_ALIAS_SETS_P (x, mem)) >> return 0; >> >> - if (nonoverlapping_memrefs_p (mem, x)) >> + if (nonoverlapping_memrefs_p (mem, x, loop_invariant)) >> return 0; >> >> The DIFFERENT_ALIAS_SETS_P would also need guarding. >> >> I think that instead of changing all present functions it would be >> better to create a new entry to the alias oracle for inter-iteration >> queries, as for example you can't distinguish true and anti dependence >> (which is also why TBAA isn't valid here). In fact the implementation >> would be the same for all of RW, WR and WW dependence types. >> >> The query would exactly match what Zdenek thinks of a >> may-alias query, so I'd suggest to simply name it may_alias_p. >> >> Richard. >> >> > Thanks >> > Bingfeng >> > >> >> -----Original Message----- >> >> From: Michael Matz [mailto:matz@suse.de] >> >> Sent: 04 August 2010 17:15 >> >> To: Bingfeng Mei >> >> Cc: Richard Guenther; Diego Novillo; gcc-patches@gcc.gnu.org >> >> Subject: RE: [PATCH] Use rtx_refs_may_alias_p instead of >> >> alias_sets_conflict_p in >> >> >> >> Hi, >> >> >> >> On Wed, 4 Aug 2010, Bingfeng Mei wrote: >> >> >> >> > Does rtx_refs_may_alias_p include offset based disambiguation >> >> (with/without >> >> > tbaa flag). >> >> >> >> Actually it doesn't (we switch off recognizing SSA names as being >> equal >> >> when called from RTL land). I forgot that this was recently changed. >> >> >> >> Unfortunately I was wrong with my assertion that TBAA can be used >> cross >> >> iteration. The catch is the new memory model, with it's asymmetric >> >> dependencies: you may move a load before a non-conflicting (type >> >> wise) store, but you may not move a store before a non-conflicting >> load. >> >> >> >> In cross-iteration situations you are posed with both cases, hence >> you >> >> can't really use type-based disambiguation (perhaps in some few >> special >> >> cases one might, but not generally). >> >> >> >> Basically for cross iteration analysis you can only use points-to >> >> information (or generally everything that is loop invariant for the >> >> loop >> >> in question, but that isn't implemented). >> >> >> >> > If yes, this patch is not correct, I need to work out a better one. >> >> >> >> It would be better to use the four different predicates we have in >> RTL >> >> land as Richard suggested, they do the right thing for each >> dependency >> >> type (switching off TBAA when appropriate). >> >> >> >> >> >> Ciao, >> >> Michael. >> > >> > > >
On Mon, Aug 9, 2010 at 5:41 AM, Bingfeng Mei <bmei@broadcom.com> wrote: > Hi, > I created a new may_alias_p function as Richard proposed. The function > disables offset-based memory disambiguation and TBAA, it is also slightly > more conservative than true/output/anti_dependence. For example, > write_dependence checks: > if (GET_MODE (x) == BLKmode && GET_CODE (XEXP (x, 0)) == SCRATCH) > return 1; > if (GET_MODE (mem) == BLKmode && GET_CODE (XEXP (mem, 0)) == SCRATCH) > return 1; > > true_depdence checks both above and following (isn't above > statement redundant then?). > if (mem_mode == BLKmode || GET_MODE (x) == BLKmode) > return 1; > may_alias_p takes more conservative check of the latter one only. > > Since insn_alias_sets_conflict_p is misnamed and only used once by ddg.c, > it & friends (walk_mems_1, walk_mems_2) are moved into ddg.c and > renamed as insns_may_alias_p. > > The patch is bootstrapped on x86_64-unknown-linux-gnu and tested. > OK for trunk? > > Thanks, > Bingfeng > > > 2010-08-10 Bingfeng Mei <bmei@broadcom.com> > > * ddg.c (walk_mems_2): Moved from alias.c, use may_alias_p instead of > alias_sets_conflict_p. > (walk_mems_1): Moved from alias.c. > (insns_may_alias_p): New function, originally insn_alias_sets_conflict_p > in alias.c. > (add_inter_loop_mem_dep): Use insns_may_alias_p now. > * cse.c (cse_insn): New argument in calling nonoverlapping_memrefs_p. > * alias.c (walk_mems_2): Moved to ddg.c. > (walk_mems_1): Ditto. > (insn_alias_sets_conflict_p): Renamed to insns_may_alias_p and moved > to ddg.c. > (nonoverlapping_memrefs_p): Add flag to guard offset-based memory > disambiguation. > *(may_alias_p): New function to check whether two memory expression > may alias or not. Currently used in buidling inter-iteration memory > dependence. > *alias.h (nonoverlapping_memrefs_p): New flag as third argument. > (insn_alias_sets_conflict_p): Removed > *rtl.h (may_alias_p): New function prototype. > Gcc is broken: ../../src-trunk/gcc/alias.c: In function 'may_alias_p': ../../src-trunk/gcc/alias.c:2537:7: error: unused variable 'ret' [-Werror=unused-variable] cc1: all warnings being treated as errors H.J.
Sorry, I will check in patch and always enable Werror next time. Cheers, Bingefng > -----Original Message----- > From: H.J. Lu [mailto:hjl.tools@gmail.com] > Sent: 09 August 2010 16:53 > To: Bingfeng Mei > Cc: Richard Guenther; Michael Matz; Diego Novillo; gcc- > patches@gcc.gnu.org > Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > alias_sets_conflict_p in > > On Mon, Aug 9, 2010 at 5:41 AM, Bingfeng Mei <bmei@broadcom.com> wrote: > > Hi, > > I created a new may_alias_p function as Richard proposed. The > function > > disables offset-based memory disambiguation and TBAA, it is also > slightly > > more conservative than true/output/anti_dependence. For example, > > write_dependence checks: > > if (GET_MODE (x) == BLKmode && GET_CODE (XEXP (x, 0)) == SCRATCH) > > return 1; > > if (GET_MODE (mem) == BLKmode && GET_CODE (XEXP (mem, 0)) == SCRATCH) > > return 1; > > > > true_depdence checks both above and following (isn't above > > statement redundant then?). > > if (mem_mode == BLKmode || GET_MODE (x) == BLKmode) > > return 1; > > may_alias_p takes more conservative check of the latter one only. > > > > Since insn_alias_sets_conflict_p is misnamed and only used once by > ddg.c, > > it & friends (walk_mems_1, walk_mems_2) are moved into ddg.c and > > renamed as insns_may_alias_p. > > > > The patch is bootstrapped on x86_64-unknown-linux-gnu and tested. > > OK for trunk? > > > > Thanks, > > Bingfeng > > > > > > 2010-08-10 Bingfeng Mei <bmei@broadcom.com> > > > > * ddg.c (walk_mems_2): Moved from alias.c, use may_alias_p > instead of > > alias_sets_conflict_p. > > (walk_mems_1): Moved from alias.c. > > (insns_may_alias_p): New function, originally > insn_alias_sets_conflict_p > > in alias.c. > > (add_inter_loop_mem_dep): Use insns_may_alias_p now. > > * cse.c (cse_insn): New argument in calling > nonoverlapping_memrefs_p. > > * alias.c (walk_mems_2): Moved to ddg.c. > > (walk_mems_1): Ditto. > > (insn_alias_sets_conflict_p): Renamed to insns_may_alias_p and > moved > > to ddg.c. > > (nonoverlapping_memrefs_p): Add flag to guard offset-based > memory > > disambiguation. > > *(may_alias_p): New function to check whether two memory > expression > > may alias or not. Currently used in buidling inter-iteration > memory > > dependence. > > *alias.h (nonoverlapping_memrefs_p): New flag as third > argument. > > (insn_alias_sets_conflict_p): Removed > > *rtl.h (may_alias_p): New function prototype. > > > > Gcc is broken: > > ../../src-trunk/gcc/alias.c: In function 'may_alias_p': > ../../src-trunk/gcc/alias.c:2537:7: error: unused variable 'ret' > [-Werror=unused-variable] > cc1: all warnings being treated as errors > > > H.J.
On Mon, Aug 9, 2010 at 5:56 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > Sorry, I will check in patch and always enable Werror next time. -Werror is enabled by default when you bootstrap, so I wonder how you could miss this? Richard. > Cheers, > Bingefng > >> -----Original Message----- >> From: H.J. Lu [mailto:hjl.tools@gmail.com] >> Sent: 09 August 2010 16:53 >> To: Bingfeng Mei >> Cc: Richard Guenther; Michael Matz; Diego Novillo; gcc- >> patches@gcc.gnu.org >> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of >> alias_sets_conflict_p in >> >> On Mon, Aug 9, 2010 at 5:41 AM, Bingfeng Mei <bmei@broadcom.com> wrote: >> > Hi, >> > I created a new may_alias_p function as Richard proposed. The >> function >> > disables offset-based memory disambiguation and TBAA, it is also >> slightly >> > more conservative than true/output/anti_dependence. For example, >> > write_dependence checks: >> > if (GET_MODE (x) == BLKmode && GET_CODE (XEXP (x, 0)) == SCRATCH) >> > return 1; >> > if (GET_MODE (mem) == BLKmode && GET_CODE (XEXP (mem, 0)) == SCRATCH) >> > return 1; >> > >> > true_depdence checks both above and following (isn't above >> > statement redundant then?). >> > if (mem_mode == BLKmode || GET_MODE (x) == BLKmode) >> > return 1; >> > may_alias_p takes more conservative check of the latter one only. >> > >> > Since insn_alias_sets_conflict_p is misnamed and only used once by >> ddg.c, >> > it & friends (walk_mems_1, walk_mems_2) are moved into ddg.c and >> > renamed as insns_may_alias_p. >> > >> > The patch is bootstrapped on x86_64-unknown-linux-gnu and tested. >> > OK for trunk? >> > >> > Thanks, >> > Bingfeng >> > >> > >> > 2010-08-10 Bingfeng Mei <bmei@broadcom.com> >> > >> > * ddg.c (walk_mems_2): Moved from alias.c, use may_alias_p >> instead of >> > alias_sets_conflict_p. >> > (walk_mems_1): Moved from alias.c. >> > (insns_may_alias_p): New function, originally >> insn_alias_sets_conflict_p >> > in alias.c. >> > (add_inter_loop_mem_dep): Use insns_may_alias_p now. >> > * cse.c (cse_insn): New argument in calling >> nonoverlapping_memrefs_p. >> > * alias.c (walk_mems_2): Moved to ddg.c. >> > (walk_mems_1): Ditto. >> > (insn_alias_sets_conflict_p): Renamed to insns_may_alias_p and >> moved >> > to ddg.c. >> > (nonoverlapping_memrefs_p): Add flag to guard offset-based >> memory >> > disambiguation. >> > *(may_alias_p): New function to check whether two memory >> expression >> > may alias or not. Currently used in buidling inter-iteration >> memory >> > dependence. >> > *alias.h (nonoverlapping_memrefs_p): New flag as third >> argument. >> > (insn_alias_sets_conflict_p): Removed >> > *rtl.h (may_alias_p): New function prototype. >> > >> >> Gcc is broken: >> >> ../../src-trunk/gcc/alias.c: In function 'may_alias_p': >> ../../src-trunk/gcc/alias.c:2537:7: error: unused variable 'ret' >> [-Werror=unused-variable] >> cc1: all warnings being treated as errors >> >> >> H.J. > > >
My tree is built together with gold, which has some warnings. So I have to disable it explicitly (not sure if it is fixed now). Bingfeng > -----Original Message----- > From: Richard Guenther [mailto:richard.guenther@gmail.com] > Sent: 09 August 2010 17:00 > To: Bingfeng Mei > Cc: H.J. Lu; Michael Matz; Diego Novillo; gcc-patches@gcc.gnu.org > Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > alias_sets_conflict_p in > > On Mon, Aug 9, 2010 at 5:56 PM, Bingfeng Mei <bmei@broadcom.com> wrote: > > Sorry, I will check in patch and always enable Werror next time. > > -Werror is enabled by default when you bootstrap, so I wonder > how you could miss this? > > Richard. > > > Cheers, > > Bingefng > > > >> -----Original Message----- > >> From: H.J. Lu [mailto:hjl.tools@gmail.com] > >> Sent: 09 August 2010 16:53 > >> To: Bingfeng Mei > >> Cc: Richard Guenther; Michael Matz; Diego Novillo; gcc- > >> patches@gcc.gnu.org > >> Subject: Re: [PATCH] Use rtx_refs_may_alias_p instead of > >> alias_sets_conflict_p in > >> > >> On Mon, Aug 9, 2010 at 5:41 AM, Bingfeng Mei <bmei@broadcom.com> > wrote: > >> > Hi, > >> > I created a new may_alias_p function as Richard proposed. The > >> function > >> > disables offset-based memory disambiguation and TBAA, it is also > >> slightly > >> > more conservative than true/output/anti_dependence. For example, > >> > write_dependence checks: > >> > if (GET_MODE (x) == BLKmode && GET_CODE (XEXP (x, 0)) == SCRATCH) > >> > return 1; > >> > if (GET_MODE (mem) == BLKmode && GET_CODE (XEXP (mem, 0)) == > SCRATCH) > >> > return 1; > >> > > >> > true_depdence checks both above and following (isn't above > >> > statement redundant then?). > >> > if (mem_mode == BLKmode || GET_MODE (x) == BLKmode) > >> > return 1; > >> > may_alias_p takes more conservative check of the latter one only. > >> > > >> > Since insn_alias_sets_conflict_p is misnamed and only used once by > >> ddg.c, > >> > it & friends (walk_mems_1, walk_mems_2) are moved into ddg.c and > >> > renamed as insns_may_alias_p. > >> > > >> > The patch is bootstrapped on x86_64-unknown-linux-gnu and tested. > >> > OK for trunk? > >> > > >> > Thanks, > >> > Bingfeng > >> > > >> > > >> > 2010-08-10 Bingfeng Mei <bmei@broadcom.com> > >> > > >> > * ddg.c (walk_mems_2): Moved from alias.c, use may_alias_p > >> instead of > >> > alias_sets_conflict_p. > >> > (walk_mems_1): Moved from alias.c. > >> > (insns_may_alias_p): New function, originally > >> insn_alias_sets_conflict_p > >> > in alias.c. > >> > (add_inter_loop_mem_dep): Use insns_may_alias_p now. > >> > * cse.c (cse_insn): New argument in calling > >> nonoverlapping_memrefs_p. > >> > * alias.c (walk_mems_2): Moved to ddg.c. > >> > (walk_mems_1): Ditto. > >> > (insn_alias_sets_conflict_p): Renamed to insns_may_alias_p > and > >> moved > >> > to ddg.c. > >> > (nonoverlapping_memrefs_p): Add flag to guard offset-based > >> memory > >> > disambiguation. > >> > *(may_alias_p): New function to check whether two memory > >> expression > >> > may alias or not. Currently used in buidling inter- > iteration > >> memory > >> > dependence. > >> > *alias.h (nonoverlapping_memrefs_p): New flag as third > >> argument. > >> > (insn_alias_sets_conflict_p): Removed > >> > *rtl.h (may_alias_p): New function prototype. > >> > > >> > >> Gcc is broken: > >> > >> ../../src-trunk/gcc/alias.c: In function 'may_alias_p': > >> ../../src-trunk/gcc/alias.c:2537:7: error: unused variable 'ret' > >> [-Werror=unused-variable] > >> cc1: all warnings being treated as errors > >> > >> > >> H.J. > > > > > >
Index: alias.c =================================================================== --- alias.c (revision 162821) +++ alias.c (working copy) @@ -454,7 +454,7 @@ { if (MEM_P (*x)) { - if (alias_sets_conflict_p (MEM_ALIAS_SET(*x), MEM_ALIAS_SET(mem))) + if (rtx_refs_may_alias_p (*x, mem, true)) return 1; return -1;
Hi, This patch uses rtx_refs_may_alias_p instead of alias_sets_conflict_p to get more accurate alias set information in walk_mems_2, which is used in building DDG. In the following example, no false cross-iteration dependence is drawn between writing to a[i] and reading b[i] in next iteration after the patch. Impact for compile-time is minimal because it only applies before modulo scheduling. Bootstrapped and tested on x86_64-unknown-linux-gnu. OK for trunk? void foo(int * restrict a, int * restrict b, int n) { int i; for(i = 0; i < n; i++) { a[i] = b[i] * 100; } } Cheers, Bingfeng Mei 2010-08-04 Bingfeng Mei <bmei@broadcom.com> * alias.c (walk_mems_2): Call rtx_refs_may_alias_p instead of alias_sets_conflict_p to get more accurate alias set information.