Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo wrote: >As discussed in the PR, Richard mentioned the method to >figure out which VAR was not set TREE_ADDRESSABLE, and >then cause this failure. It is address_expression which >build addr_expr (build_fold_addr_expr_loc), but not set >TREE_ADDRESSABLE. > >I drafted this patch with reference the comments from Richard >in this PR, while I'm not quite sure if more thing need to do. >So, please have review, thanks! > >Bootstrap and regtest pass on ppc64le. Is this ok for trunk? I suggest to use mark_addresssable unless we're sure expr is always an entity where TREE_ADDRESSABLE has the desired meaning. Richard. >Jiufu Guo. > >2021-05-14 Richard Biener > Jiufu Guo > > PR go/100537 > * go-gcc.cc > (Gcc_backend::address_expression): Set TREE_ADDRESSABLE. > >--- > gcc/go/go-gcc.cc | 1 + > 1 file changed, 1 insertion(+) > >diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc >index 5d9dbb5d068..8ed20a3b479 100644 >--- a/gcc/go/go-gcc.cc >+++ b/gcc/go/go-gcc.cc >@@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* >bexpr, Location location) > if (expr == error_mark_node) > return this->error_expression(); > >+ TREE_ADDRESSABLE (expr) = 1; > tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); > return this->make_expression(ret); > }
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On 2021-05-14 15:15, Richard Biener wrote: On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo wrote: As discussed in the PR, Richard mentioned the method to figure out which VAR was not set TREE_ADDRESSABLE, and then cause this failure. It is address_expression which build addr_expr (build_fold_addr_expr_loc), but not set TREE_ADDRESSABLE. I drafted this patch with reference the comments from Richard in this PR, while I'm not quite sure if more thing need to do. So, please have review, thanks! Bootstrap and regtest pass on ppc64le. Is this ok for trunk? I suggest to use mark_addresssable unless we're sure expr is always an entity where TREE_ADDRESSABLE has the desired meaning. I notice you mentioned "mark_addresssable" in PR. And I had tried yesterday, it cause new ICEs at gimple-expr.c:918 below line: && cfun->gimple_df != NULL Richard. Jiufu Guo. 2021-05-14 Richard Biener Jiufu Guo PR go/100537 * go-gcc.cc (Gcc_backend::address_expression): Set TREE_ADDRESSABLE. --- gcc/go/go-gcc.cc | 1 + 1 file changed, 1 insertion(+) diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..8ed20a3b479 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* bexpr, Location location) if (expr == error_mark_node) return this->error_expression(); + TREE_ADDRESSABLE (expr) = 1; tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); return this->make_expression(ret); }
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: On 2021-05-14 15:15, Richard Biener wrote: On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo wrote: As discussed in the PR, Richard mentioned the method to figure out which VAR was not set TREE_ADDRESSABLE, and then cause this failure. It is address_expression which build addr_expr (build_fold_addr_expr_loc), but not set TREE_ADDRESSABLE. I drafted this patch with reference the comments from Richard in this PR, while I'm not quite sure if more thing need to do. So, please have review, thanks! Bootstrap and regtest pass on ppc64le. Is this ok for trunk? I suggest to use mark_addresssable unless we're sure expr is always an entity where TREE_ADDRESSABLE has the desired meaning. Thanks, Richard! You point out the root concern, I'm not sure ;) With looking at code "mark_addresssable" and code around tree-ssa.c:1013, VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing TREE_ADDRESSABLE. So, just wondering if these entities need to be marked as TREE_ADDRESSABLE? diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..85d324a92cc 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* bexpr, Location location) if (expr == error_mark_node) return this->error_expression(); + if ((VAR_P(expr) + || TREE_CODE(expr) == PARM_DECL + || TREE_CODE(expr) == RESULT_DECL) +TREE_ADDRESSABLE (expr) = 1; + tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); return this->make_expression(ret); } Or call mark_addressable, and update mark_addressable to avoid NULL pointer ICE: The below patch also pass bootstrap-debug. diff --git a/gcc/gimple-expr.c b/gcc/gimple-expr.c index b8c732b632a..f682841391b 100644 --- a/gcc/gimple-expr.c +++ b/gcc/gimple-expr.c @@ -915,6 +915,7 @@ mark_addressable (tree x) if (TREE_CODE (x) == VAR_DECL && !DECL_EXTERNAL (x) && !TREE_STATIC (x) + && cfun != NULL && cfun->gimple_df != NULL && cfun->gimple_df->decls_to_pointers != NULL) { diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..fe9dfaf8579 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* bexpr, Location location) if (expr == error_mark_node) return this->error_expression(); + mark_addressable(expr); tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); return this->make_expression(ret); } I notice you mentioned "mark_addresssable" in PR. And I had tried yesterday, it cause new ICEs at gimple-expr.c:918 below line: && cfun->gimple_df != NULL Richard. Jiufu Guo. 2021-05-14 Richard Biener Jiufu Guo PR go/100537 * go-gcc.cc (Gcc_backend::address_expression): Set TREE_ADDRESSABLE. --- gcc/go/go-gcc.cc | 1 + 1 file changed, 1 insertion(+) diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..8ed20a3b479 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* bexpr, Location location) if (expr == error_mark_node) return this->error_expression(); + TREE_ADDRESSABLE (expr) = 1; tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); return this->make_expression(ret); }
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches wrote: > > On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > > On 2021-05-14 15:15, Richard Biener wrote: > >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo > >> wrote: > >>> As discussed in the PR, Richard mentioned the method to > >>> figure out which VAR was not set TREE_ADDRESSABLE, and > >>> then cause this failure. It is address_expression which > >>> build addr_expr (build_fold_addr_expr_loc), but not set > >>> TREE_ADDRESSABLE. > >>> > >>> I drafted this patch with reference the comments from Richard > >>> in this PR, while I'm not quite sure if more thing need to do. > >>> So, please have review, thanks! > >>> > >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? > >> > >> I suggest to use mark_addresssable unless we're sure expr is always an > >> entity where TREE_ADDRESSABLE has the desired meaning. > > Thanks, Richard! > You point out the root concern, I'm not sure ;) > > With looking at code "mark_addresssable" and code around > tree-ssa.c:1013, > VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing > TREE_ADDRESSABLE. > So, just wondering if these entities need to be marked as > TREE_ADDRESSABLE? > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > index 5d9dbb5d068..85d324a92cc 100644 > --- a/gcc/go/go-gcc.cc > +++ b/gcc/go/go-gcc.cc > @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* > bexpr, Location location) > if (expr == error_mark_node) > return this->error_expression(); > > + if ((VAR_P(expr) > + || TREE_CODE(expr) == PARM_DECL > + || TREE_CODE(expr) == RESULT_DECL) > +TREE_ADDRESSABLE (expr) = 1; > + The root concern is that mark_addressable does while (handled_component_p (x)) x = TREE_OPERAND (x, 0); and I do not know the constraints on 'expr' as passed to Gcc_backend::address_expression. I think we need input from Ian here. Most FEs have their own *_mark_addressable function where they also emit diagnostics (guess this is handled in the actual Go frontend). Since Gcc_backend does lowering to GENERIC using a middle-end is probably OK. > tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); > return this->make_expression(ret); > } > > > Or call mark_addressable, and update mark_addressable to avoid NULL > pointer ICE: > The below patch also pass bootstrap-debug. > > diff --git a/gcc/gimple-expr.c b/gcc/gimple-expr.c > index b8c732b632a..f682841391b 100644 > --- a/gcc/gimple-expr.c > +++ b/gcc/gimple-expr.c > @@ -915,6 +915,7 @@ mark_addressable (tree x) > if (TREE_CODE (x) == VAR_DECL > && !DECL_EXTERNAL (x) > && !TREE_STATIC (x) > + && cfun != NULL I'd be OK with this hunk of course. > && cfun->gimple_df != NULL > && cfun->gimple_df->decls_to_pointers != NULL) > { > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > index 5d9dbb5d068..fe9dfaf8579 100644 > --- a/gcc/go/go-gcc.cc > +++ b/gcc/go/go-gcc.cc > @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* > bexpr, Location location) > if (expr == error_mark_node) > return this->error_expression(); > > + mark_addressable(expr); > tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); > return this->make_expression(ret); > } > > > > > > I notice you mentioned "mark_addresssable" in PR. > > And I had tried yesterday, it cause new ICEs at gimple-expr.c:918 > > below line: > > > > && cfun->gimple_df != NULL > > > > > > > >> > >> Richard. > >> > >>> Jiufu Guo. > >>> > >>> 2021-05-14 Richard Biener > >>> Jiufu Guo > >>> > >>> PR go/100537 > >>> * go-gcc.cc > >>> (Gcc_backend::address_expression): Set TREE_ADDRESSABLE. > >>> > >>> --- > >>> gcc/go/go-gcc.cc | 1 + > >>> 1 file changed, 1 insertion(+) > >>> > >>> diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > >>> index 5d9dbb5d068..8ed20a3b479 100644 > >>> --- a/gcc/go/go-gcc.cc > >>> +++ b/gcc/go/go-gcc.cc > >>> @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* > >>> bexpr, Location location) > >>> if (expr == error_mark_node) > >>> return this->error_expression(); > >>> > >>> + TREE_ADDRESSABLE (expr) = 1; > >>> tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); > >>> return this->make_expression(ret); > >>> }
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On 2021-05-17 16:17, Richard Biener wrote: On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches wrote: On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > On 2021-05-14 15:15, Richard Biener wrote: >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo >> wrote: >>> As discussed in the PR, Richard mentioned the method to >>> figure out which VAR was not set TREE_ADDRESSABLE, and >>> then cause this failure. It is address_expression which >>> build addr_expr (build_fold_addr_expr_loc), but not set >>> TREE_ADDRESSABLE. >>> >>> I drafted this patch with reference the comments from Richard >>> in this PR, while I'm not quite sure if more thing need to do. >>> So, please have review, thanks! >>> >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? >> >> I suggest to use mark_addresssable unless we're sure expr is always an >> entity where TREE_ADDRESSABLE has the desired meaning. Thanks, Richard! You point out the root concern, I'm not sure ;) With looking at code "mark_addresssable" and code around tree-ssa.c:1013, VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing TREE_ADDRESSABLE. So, just wondering if these entities need to be marked as TREE_ADDRESSABLE? diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..85d324a92cc 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* bexpr, Location location) if (expr == error_mark_node) return this->error_expression(); + if ((VAR_P(expr) + || TREE_CODE(expr) == PARM_DECL + || TREE_CODE(expr) == RESULT_DECL) +TREE_ADDRESSABLE (expr) = 1; + The root concern is that mark_addressable does while (handled_component_p (x)) x = TREE_OPERAND (x, 0); and I do not know the constraints on 'expr' as passed to Gcc_backend::address_expression. I think we need input from Ian here. Most FEs have their own *_mark_addressable function where they also emit diagnostics (guess this is handled in the actual Go frontend). Since Gcc_backend does lowering to GENERIC using a middle-end is probably OK. Yeap. Hope this patch is ok, then the bootstrap could pass. Otherwise, we may need more help from Ian and guys ;) Jiufu Guo. tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); return this->make_expression(ret); } Or call mark_addressable, and update mark_addressable to avoid NULL pointer ICE: The below patch also pass bootstrap-debug. diff --git a/gcc/gimple-expr.c b/gcc/gimple-expr.c index b8c732b632a..f682841391b 100644 --- a/gcc/gimple-expr.c +++ b/gcc/gimple-expr.c @@ -915,6 +915,7 @@ mark_addressable (tree x) if (TREE_CODE (x) == VAR_DECL && !DECL_EXTERNAL (x) && !TREE_STATIC (x) + && cfun != NULL I'd be OK with this hunk of course. && cfun->gimple_df != NULL && cfun->gimple_df->decls_to_pointers != NULL) { diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..fe9dfaf8579 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* bexpr, Location location) if (expr == error_mark_node) return this->error_expression(); + mark_addressable(expr); tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); return this->make_expression(ret); } > > I notice you mentioned "mark_addresssable" in PR. > And I had tried yesterday, it cause new ICEs at gimple-expr.c:918 > below line: > > && cfun->gimple_df != NULL > > > >> >> Richard. >> >>> Jiufu Guo. >>> >>> 2021-05-14 Richard Biener >>> Jiufu Guo >>> >>> PR go/100537 >>> * go-gcc.cc >>> (Gcc_backend::address_expression): Set TREE_ADDRESSABLE. >>> >>> --- >>> gcc/go/go-gcc.cc | 1 + >>> 1 file changed, 1 insertion(+) >>> >>> diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc >>> index 5d9dbb5d068..8ed20a3b479 100644 >>> --- a/gcc/go/go-gcc.cc >>> +++ b/gcc/go/go-gcc.cc >>> @@ -1680,6 +1680,7 @@ Gcc_backend::address_expression(Bexpression* >>> bexpr, Location location) >>> if (expr == error_mark_node) >>> return this->error_expression(); >>> >>> + TREE_ADDRESSABLE (expr) = 1; >>> tree ret = build_fold_addr_expr_loc(location.gcc_location(), expr); >>> return this->make_expression(ret); >>> }
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On Mon, May 17, 2021 at 1:17 AM Richard Biener via Gcc-patches wrote: > > On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches > wrote: > > > > On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > > > On 2021-05-14 15:15, Richard Biener wrote: > > >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo > > >> wrote: > > >>> As discussed in the PR, Richard mentioned the method to > > >>> figure out which VAR was not set TREE_ADDRESSABLE, and > > >>> then cause this failure. It is address_expression which > > >>> build addr_expr (build_fold_addr_expr_loc), but not set > > >>> TREE_ADDRESSABLE. > > >>> > > >>> I drafted this patch with reference the comments from Richard > > >>> in this PR, while I'm not quite sure if more thing need to do. > > >>> So, please have review, thanks! > > >>> > > >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? > > >> > > >> I suggest to use mark_addresssable unless we're sure expr is always an > > >> entity where TREE_ADDRESSABLE has the desired meaning. > > > > Thanks, Richard! > > You point out the root concern, I'm not sure ;) > > > > With looking at code "mark_addresssable" and code around > > tree-ssa.c:1013, > > VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing > > TREE_ADDRESSABLE. > > So, just wondering if these entities need to be marked as > > TREE_ADDRESSABLE? > > > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > > index 5d9dbb5d068..85d324a92cc 100644 > > --- a/gcc/go/go-gcc.cc > > +++ b/gcc/go/go-gcc.cc > > @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* > > bexpr, Location location) > > if (expr == error_mark_node) > > return this->error_expression(); > > > > + if ((VAR_P(expr) > > + || TREE_CODE(expr) == PARM_DECL > > + || TREE_CODE(expr) == RESULT_DECL) > > +TREE_ADDRESSABLE (expr) = 1; > > + > > The root concern is that mark_addressable does > > while (handled_component_p (x)) > x = TREE_OPERAND (x, 0); > > and I do not know the constraints on 'expr' as passed to > Gcc_backend::address_expression. > > I think we need input from Ian here. Most FEs have their own > *_mark_addressable > function where they also emit diagnostics (guess this is handled in > the actual Go frontend). > Since Gcc_backend does lowering to GENERIC using a middle-end is probably OK. I doubt I understand all the issues here. In general the Go frontend only takes the addresses of VAR_DECLs or PARM_DECLs. It doesn't bother to set TREE_ADDRESSABLE for global variables for which TREE_STATIC or DECL_EXTERNAL is true. For local variables it sets TREE_ADDRESSABLE based on the is_address_taken parameter to Gcc_backend::local_variable, and similarly for PARM_DECLs and Gcc_backend::parameter_variable. The name in the bug report is for a string initializer, which should be TREE_STATIC == 1 and TREE_PUBLIC == 0. Perhaps the fix is simply to set TREE_ADDRESSABLE in Gcc_backend::immutable_struct and Gcc_backend::implicit_variable. I can't see how it would hurt to set TREE_ADDRESSABLE unnecessarily for a TREE_STATIC variable. But, again, I doubt I understand all the issues here. Ian
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On Mon, 17 May 2021, Ian Lance Taylor wrote: > On Mon, May 17, 2021 at 1:17 AM Richard Biener via Gcc-patches > wrote: > > > > On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches > > wrote: > > > > > > On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > > > > On 2021-05-14 15:15, Richard Biener wrote: > > > >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo > > > >> wrote: > > > >>> As discussed in the PR, Richard mentioned the method to > > > >>> figure out which VAR was not set TREE_ADDRESSABLE, and > > > >>> then cause this failure. It is address_expression which > > > >>> build addr_expr (build_fold_addr_expr_loc), but not set > > > >>> TREE_ADDRESSABLE. > > > >>> > > > >>> I drafted this patch with reference the comments from Richard > > > >>> in this PR, while I'm not quite sure if more thing need to do. > > > >>> So, please have review, thanks! > > > >>> > > > >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? > > > >> > > > >> I suggest to use mark_addresssable unless we're sure expr is always an > > > >> entity where TREE_ADDRESSABLE has the desired meaning. > > > > > > Thanks, Richard! > > > You point out the root concern, I'm not sure ;) > > > > > > With looking at code "mark_addresssable" and code around > > > tree-ssa.c:1013, > > > VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing > > > TREE_ADDRESSABLE. > > > So, just wondering if these entities need to be marked as > > > TREE_ADDRESSABLE? > > > > > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > > > index 5d9dbb5d068..85d324a92cc 100644 > > > --- a/gcc/go/go-gcc.cc > > > +++ b/gcc/go/go-gcc.cc > > > @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* > > > bexpr, Location location) > > > if (expr == error_mark_node) > > > return this->error_expression(); > > > > > > + if ((VAR_P(expr) > > > + || TREE_CODE(expr) == PARM_DECL > > > + || TREE_CODE(expr) == RESULT_DECL) > > > +TREE_ADDRESSABLE (expr) = 1; > > > + > > > > The root concern is that mark_addressable does > > > > while (handled_component_p (x)) > > x = TREE_OPERAND (x, 0); > > > > and I do not know the constraints on 'expr' as passed to > > Gcc_backend::address_expression. > > > > I think we need input from Ian here. Most FEs have their own > > *_mark_addressable > > function where they also emit diagnostics (guess this is handled in > > the actual Go frontend). > > Since Gcc_backend does lowering to GENERIC using a middle-end is probably > > OK. > > I doubt I understand all the issues here. > > In general the Go frontend only takes the addresses of VAR_DECLs or > PARM_DECLs. It doesn't bother to set TREE_ADDRESSABLE for global > variables for which TREE_STATIC or DECL_EXTERNAL is true. For local > variables it sets TREE_ADDRESSABLE based on the is_address_taken > parameter to Gcc_backend::local_variable, and similarly for PARM_DECLs > and Gcc_backend::parameter_variable. > > The name in the bug report is for a string initializer, which should > be TREE_STATIC == 1 and TREE_PUBLIC == 0. Perhaps the fix is simply > to set TREE_ADDRESSABLE in Gcc_backend::immutable_struct and > Gcc_backend::implicit_variable. I can't see how it would hurt to set > TREE_ADDRESSABLE unnecessarily for a TREE_STATIC variable. > > But, again, I doubt I understand all the issues here. GENERIC requires TREE_ADDRESSABLE to be set on all address-taken VAR_DECLs, PARM_DECLs and RESULT_DECLs - the gimplifier is the first to require this for correctness. Setting TREE_ADDRESSABLE when the address is not taken is harmless and at most results in missed optimizations (on most entities we are able to clear the flag later). We're currently quite forgiving with this though (still the gimplifier can generate wrong-code). The trigger of the current failure removed one "forgiveness", I do plan to remove a few more. guojiufu's patch works for me but as said I'm not sure if there's a better place to set TREE_ADDRESSABLE for entities that have their address taken - definitely catching the places where you build an ADDR_EXPR are the most obvious ones. Richard.
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On 2021-05-18 14:58, Richard Biener wrote: On Mon, 17 May 2021, Ian Lance Taylor wrote: On Mon, May 17, 2021 at 1:17 AM Richard Biener via Gcc-patches wrote: > > On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches > wrote: > > > > On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > > > On 2021-05-14 15:15, Richard Biener wrote: > > >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo > > >> wrote: > > >>> As discussed in the PR, Richard mentioned the method to > > >>> figure out which VAR was not set TREE_ADDRESSABLE, and > > >>> then cause this failure. It is address_expression which > > >>> build addr_expr (build_fold_addr_expr_loc), but not set > > >>> TREE_ADDRESSABLE. > > >>> > > >>> I drafted this patch with reference the comments from Richard > > >>> in this PR, while I'm not quite sure if more thing need to do. > > >>> So, please have review, thanks! > > >>> > > >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? > > >> > > >> I suggest to use mark_addresssable unless we're sure expr is always an > > >> entity where TREE_ADDRESSABLE has the desired meaning. > > > > Thanks, Richard! > > You point out the root concern, I'm not sure ;) > > > > With looking at code "mark_addresssable" and code around > > tree-ssa.c:1013, > > VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing > > TREE_ADDRESSABLE. > > So, just wondering if these entities need to be marked as > > TREE_ADDRESSABLE? > > > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > > index 5d9dbb5d068..85d324a92cc 100644 > > --- a/gcc/go/go-gcc.cc > > +++ b/gcc/go/go-gcc.cc > > @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* > > bexpr, Location location) > > if (expr == error_mark_node) > > return this->error_expression(); > > > > + if ((VAR_P(expr) > > + || TREE_CODE(expr) == PARM_DECL > > + || TREE_CODE(expr) == RESULT_DECL) > > +TREE_ADDRESSABLE (expr) = 1; > > + > > The root concern is that mark_addressable does > > while (handled_component_p (x)) > x = TREE_OPERAND (x, 0); > > and I do not know the constraints on 'expr' as passed to > Gcc_backend::address_expression. > > I think we need input from Ian here. Most FEs have their own *_mark_addressable > function where they also emit diagnostics (guess this is handled in > the actual Go frontend). > Since Gcc_backend does lowering to GENERIC using a middle-end is probably OK. I doubt I understand all the issues here. In general the Go frontend only takes the addresses of VAR_DECLs or PARM_DECLs. It doesn't bother to set TREE_ADDRESSABLE for global variables for which TREE_STATIC or DECL_EXTERNAL is true. For local variables it sets TREE_ADDRESSABLE based on the is_address_taken parameter to Gcc_backend::local_variable, and similarly for PARM_DECLs and Gcc_backend::parameter_variable. The name in the bug report is for a string initializer, which should be TREE_STATIC == 1 and TREE_PUBLIC == 0. Perhaps the fix is simply to set TREE_ADDRESSABLE in Gcc_backend::immutable_struct and Gcc_backend::implicit_variable. I can't see how it would hurt to set TREE_ADDRESSABLE unnecessarily for a TREE_STATIC variable. One more finding: Gcc_backend::implicit_variable -> build_decl is called for "" at Unary_expression::do_get_backend (expressions.cc:5322). And, this code (as below) from "expressions.cc:5322" Unary_expression::do_get_backend (expressions.cc:5322): gogo->backend()->implicit_variable(var_name, "", btype, true, true, false, 0); where var_name is go..C479 This code is under **"case OPERATOR_AND:"** of a switch statement. Unary_expression with OPERATOR_AND is "&" expression, I guess, it may look as taking address. And as the log mentioned: "PHI ". In this phi, &go..C479(59) would be the Unary_expression with OPERATOR_AND on "go..C479". So, I guess, we may able to treat "Unary_expression with OPERATOR_AND" as address_taken operation. Then it would be necessary to mark addressable. address_expression: "ret = gogo->backend()->address_expression(bexpr, loc);" (expressions.cc:5330) is already called under "Unary_expression with OPERATOR_AND". Does this make sense? If so, we may set "TREE_ADDRESSABLE" just before expressions.cc:5330? Hope this finding is helpful. BR. Jiufu Guo. But, again, I doubt I understand all the issues here. GENERIC requires TREE_ADDRESSABLE to be set on all address-taken VAR_DECLs, PARM_DECLs and RESULT_DECLs - the gimplifier is the first to require this for correctness. Setting TREE_ADDRESSABLE when the address is not taken is harmless and at most results in missed optimizations (on most entities we are able to clear the flag later). We're currently quite forgiving with this though (still the gimplifier can generate wrong-code). The trigger of the current failure removed one "forgiveness", I do plan to remove a few more. guojiufu's patch works for me but as said I'm not sure if there's a better place to set TREE_ADDRESSABLE for entities that have
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
On 2021-05-18 14:58, Richard Biener wrote: On Mon, 17 May 2021, Ian Lance Taylor wrote: On Mon, May 17, 2021 at 1:17 AM Richard Biener via Gcc-patches wrote: > > On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches > wrote: > > > > On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > > > On 2021-05-14 15:15, Richard Biener wrote: > > >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo > > >> wrote: > > >>> As discussed in the PR, Richard mentioned the method to > > >>> figure out which VAR was not set TREE_ADDRESSABLE, and > > >>> then cause this failure. It is address_expression which > > >>> build addr_expr (build_fold_addr_expr_loc), but not set > > >>> TREE_ADDRESSABLE. > > >>> > > >>> I drafted this patch with reference the comments from Richard > > >>> in this PR, while I'm not quite sure if more thing need to do. > > >>> So, please have review, thanks! > > >>> > > >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? > > >> > > >> I suggest to use mark_addresssable unless we're sure expr is always an > > >> entity where TREE_ADDRESSABLE has the desired meaning. > > > > Thanks, Richard! > > You point out the root concern, I'm not sure ;) > > > > With looking at code "mark_addresssable" and code around > > tree-ssa.c:1013, > > VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing > > TREE_ADDRESSABLE. > > So, just wondering if these entities need to be marked as > > TREE_ADDRESSABLE? > > > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > > index 5d9dbb5d068..85d324a92cc 100644 > > --- a/gcc/go/go-gcc.cc > > +++ b/gcc/go/go-gcc.cc > > @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* > > bexpr, Location location) > > if (expr == error_mark_node) > > return this->error_expression(); > > > > + if ((VAR_P(expr) > > + || TREE_CODE(expr) == PARM_DECL > > + || TREE_CODE(expr) == RESULT_DECL) > > +TREE_ADDRESSABLE (expr) = 1; > > + > > The root concern is that mark_addressable does > > while (handled_component_p (x)) > x = TREE_OPERAND (x, 0); > > and I do not know the constraints on 'expr' as passed to > Gcc_backend::address_expression. > > I think we need input from Ian here. Most FEs have their own *_mark_addressable > function where they also emit diagnostics (guess this is handled in > the actual Go frontend). > Since Gcc_backend does lowering to GENERIC using a middle-end is probably OK. I doubt I understand all the issues here. In general the Go frontend only takes the addresses of VAR_DECLs or PARM_DECLs. It doesn't bother to set TREE_ADDRESSABLE for global variables for which TREE_STATIC or DECL_EXTERNAL is true. For local variables it sets TREE_ADDRESSABLE based on the is_address_taken parameter to Gcc_backend::local_variable, and similarly for PARM_DECLs and Gcc_backend::parameter_variable. The name in the bug report is for a string initializer, which should be TREE_STATIC == 1 and TREE_PUBLIC == 0. Perhaps the fix is simply to set TREE_ADDRESSABLE in Gcc_backend::immutable_struct and Gcc_backend::implicit_variable. I can't see how it would hurt to set TREE_ADDRESSABLE unnecessarily for a TREE_STATIC variable. But, again, I doubt I understand all the issues here. GENERIC requires TREE_ADDRESSABLE to be set on all address-taken VAR_DECLs, PARM_DECLs and RESULT_DECLs - the gimplifier is the first to require this for correctness. Setting TREE_ADDRESSABLE when the address is not taken is harmless and at most results in missed optimizations (on most entities we are able to clear the flag later). We're currently quite forgiving with this though (still the gimplifier can generate wrong-code). The trigger of the current failure removed one "forgiveness", I do plan to remove a few more. guojiufu's patch works for me but as said I'm not sure if there's a better place to set TREE_ADDRESSABLE for entities that have their address taken - definitely catching the places where you build an ADDR_EXPR are the most obvious ones. Richard. I tested below patch As Ian said, bootstrap pass. diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc index 5d9dbb5d068..529f657598a 100644 --- a/gcc/go/go-gcc.cc +++ b/gcc/go/go-gcc.cc @@ -2943,6 +2943,7 @@ Gcc_backend::implicit_variable(const std::string& name, TREE_STATIC(decl) = 1; TREE_USED(decl) = 1; DECL_ARTIFICIAL(decl) = 1; + TREE_ADDRESSABLE(decl) = 1; if (is_common) { DECL_COMMON(decl) = 1; @@ -3053,6 +3054,7 @@ Gcc_backend::immutable_struct(const std::string& name, TREE_READONLY(decl) = 1; TREE_CONSTANT(decl) = 1; DECL_ARTIFICIAL(decl) = 1; + TREE_ADDRESSABLE(decl) = 1; if (!is_hidden) TREE_PUBLIC(decl) = 1; if (! asm_name.empty()) While, I hacked a patch for OPERATOR_AND. I'm thinking it may be acceptable. BR. Jiufu Guo. Bootstrap failure go [PR100537] In general the Go frontend only takes the
Re: [PATCH] go/100537 - Bootstrap-O3 and bootstrap-debug fail
I've committed a patch to the Go frontend that fixes this problem. Thanks for the analysis. Ian On Thu, May 20, 2021 at 1:59 AM guojiufu wrote: > > On 2021-05-18 14:58, Richard Biener wrote: > > On Mon, 17 May 2021, Ian Lance Taylor wrote: > > > >> On Mon, May 17, 2021 at 1:17 AM Richard Biener via Gcc-patches > >> wrote: > >> > > >> > On Fri, May 14, 2021 at 11:19 AM guojiufu via Gcc-patches > >> > wrote: > >> > > > >> > > On 2021-05-14 15:39, guojiufu via Gcc-patches wrote: > >> > > > On 2021-05-14 15:15, Richard Biener wrote: > >> > > >> On May 14, 2021 4:52:56 AM GMT+02:00, Jiufu Guo > >> > > >> wrote: > >> > > >>> As discussed in the PR, Richard mentioned the method to > >> > > >>> figure out which VAR was not set TREE_ADDRESSABLE, and > >> > > >>> then cause this failure. It is address_expression which > >> > > >>> build addr_expr (build_fold_addr_expr_loc), but not set > >> > > >>> TREE_ADDRESSABLE. > >> > > >>> > >> > > >>> I drafted this patch with reference the comments from Richard > >> > > >>> in this PR, while I'm not quite sure if more thing need to do. > >> > > >>> So, please have review, thanks! > >> > > >>> > >> > > >>> Bootstrap and regtest pass on ppc64le. Is this ok for trunk? > >> > > >> > >> > > >> I suggest to use mark_addresssable unless we're sure expr is always > >> > > >> an > >> > > >> entity where TREE_ADDRESSABLE has the desired meaning. > >> > > > >> > > Thanks, Richard! > >> > > You point out the root concern, I'm not sure ;) > >> > > > >> > > With looking at code "mark_addresssable" and code around > >> > > tree-ssa.c:1013, > >> > > VAR_P, PARM_DECL, and RESULT_DECL are checked before accessing > >> > > TREE_ADDRESSABLE. > >> > > So, just wondering if these entities need to be marked as > >> > > TREE_ADDRESSABLE? > >> > > > >> > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > >> > > index 5d9dbb5d068..85d324a92cc 100644 > >> > > --- a/gcc/go/go-gcc.cc > >> > > +++ b/gcc/go/go-gcc.cc > >> > > @@ -1680,6 +1680,11 @@ Gcc_backend::address_expression(Bexpression* > >> > > bexpr, Location location) > >> > > if (expr == error_mark_node) > >> > > return this->error_expression(); > >> > > > >> > > + if ((VAR_P(expr) > >> > > + || TREE_CODE(expr) == PARM_DECL > >> > > + || TREE_CODE(expr) == RESULT_DECL) > >> > > +TREE_ADDRESSABLE (expr) = 1; > >> > > + > >> > > >> > The root concern is that mark_addressable does > >> > > >> > while (handled_component_p (x)) > >> > x = TREE_OPERAND (x, 0); > >> > > >> > and I do not know the constraints on 'expr' as passed to > >> > Gcc_backend::address_expression. > >> > > >> > I think we need input from Ian here. Most FEs have their own > >> > *_mark_addressable > >> > function where they also emit diagnostics (guess this is handled in > >> > the actual Go frontend). > >> > Since Gcc_backend does lowering to GENERIC using a middle-end is > >> > probably OK. > >> > >> I doubt I understand all the issues here. > >> > >> In general the Go frontend only takes the addresses of VAR_DECLs or > >> PARM_DECLs. It doesn't bother to set TREE_ADDRESSABLE for global > >> variables for which TREE_STATIC or DECL_EXTERNAL is true. For local > >> variables it sets TREE_ADDRESSABLE based on the is_address_taken > >> parameter to Gcc_backend::local_variable, and similarly for PARM_DECLs > >> and Gcc_backend::parameter_variable. > >> > >> The name in the bug report is for a string initializer, which should > >> be TREE_STATIC == 1 and TREE_PUBLIC == 0. Perhaps the fix is simply > >> to set TREE_ADDRESSABLE in Gcc_backend::immutable_struct and > >> Gcc_backend::implicit_variable. I can't see how it would hurt to set > >> TREE_ADDRESSABLE unnecessarily for a TREE_STATIC variable. > >> > >> But, again, I doubt I understand all the issues here. > > > > GENERIC requires TREE_ADDRESSABLE to be set on all address-taken > > VAR_DECLs, PARM_DECLs and RESULT_DECLs - the gimplifier is the > > first to require this for correctness. Setting TREE_ADDRESSABLE > > when the address is not taken is harmless and at most results in > > missed optimizations (on most entities we are able to clear the > > flag later). > > > > We're currently quite forgiving with this though (still the > > gimplifier can generate wrong-code). The trigger of the current > > failure removed one "forgiveness", I do plan to remove a few more. > > > > guojiufu's patch works for me but as said I'm not sure if there's > > a better place to set TREE_ADDRESSABLE for entities that have > > their address taken - definitely catching the places where > > you build an ADDR_EXPR are the most obvious ones. > > > > Richard. > > I tested below patch As Ian said, bootstrap pass. > > > diff --git a/gcc/go/go-gcc.cc b/gcc/go/go-gcc.cc > index 5d9dbb5d068..529f657598a 100644 > --- a/gcc/go/go-gcc.cc > +++ b/gcc/go/go-gcc.cc > @@ -2943,6 +2943,7 @@ Gcc_backend::implicit_variable(const std::string& > name, > TREE_STATIC(decl) = 1; >