* [PATCH][ARM] PR 65694: Properly sign-extend large numbers before passing to GEN_INT in arm_canonicalize_comparison @ 2015-04-10 8:33 Kyrill Tkachov 2015-04-10 8:57 ` Jakub Jelinek 0 siblings, 1 reply; 4+ messages in thread From: Kyrill Tkachov @ 2015-04-10 8:33 UTC (permalink / raw) To: GCC Patches; +Cc: Ramana Radhakrishnan, Richard Earnshaw [-- Attachment #1: Type: text/plain, Size: 1552 bytes --] Hi all, This ICE has a similar cause to PR 64600. The arm backend creates a const_int with a large value and doesn't properly sign-extend it. If that rtx then happens to pass through the simplify-rtx machinery and ends up in wide-int land the asserts there catch it and ICE. In this case it was 0x7fffffff having 1 added to it to make 0x80000000 which should have been sign-extended to 0xffffffff80000000. The code that caused this is: class G { public: void allocate(int p1) { if (p1 > max_size()) operator new(sizeof(short)); } unsigned max_size() { return -1 / sizeof(short); } }; but it also needs to go through simplify-rtx to ICE, so the testcase contains the context as well. Similarly, the solution here is to use ARM_SIGN_EXTEND on the SImode constants and trunc_int_for_mode for the DImode ones. Bootstrapped and tested on arm-none-linux-gnueabihf. Built SPEC2006 with it. No code changes anywhere except in one instruction where a: cmp r6, #2147483648 is transformed into a: cmp r6, #-2147483648 Both assemble to the same thing, the disassembly is: cmp.w r6, #2147483648 ; 0x80000000 Ok for trunk? Thanks, Kyrill 2015-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com> * config/arm/arm.c (arm_canonicalize_comparison): Use ARM_SIGN_EXTEND when creating +1 values for SImode and trunc_int_for_mode for similar DImode operations. 2015-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com> * g++.dg/torture/pr65694.C: New test. [-- Warning: decoded text below may be mangled, UTF-8 assumed --] [-- Attachment #2: arm-extend-cmp.patch --] [-- Type: text/x-patch; name=arm-extend-cmp.patch, Size: 6394 bytes --] commit 90fbcc1f74efd4b5f077fa6caea91d24548cca34 Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com> Date: Wed Apr 8 17:08:17 2015 +0100 [ARM] PR target/65694: Use ARM_SIGN_EXTEND in arm_canonicalize_comparison diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c index 369cb67..5342b33 100644 --- a/gcc/config/arm/arm.c +++ b/gcc/config/arm/arm.c @@ -4984,7 +4984,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, if (i != maxval && arm_const_double_by_immediates (GEN_INT (i + 1))) { - *op1 = GEN_INT (i + 1); + *op1 = GEN_INT (trunc_int_for_mode (i + 1, DImode)); *code = *code == GT ? GE : LT; return; } @@ -4994,7 +4994,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, if (i != ~((unsigned HOST_WIDE_INT) 0) && arm_const_double_by_immediates (GEN_INT (i + 1))) { - *op1 = GEN_INT (i + 1); + *op1 = GEN_INT (trunc_int_for_mode (i + 1, DImode)); *code = *code == GTU ? GEU : LTU; return; } @@ -5047,7 +5047,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, if (i != maxval && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) { - *op1 = GEN_INT (i + 1); + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); *code = *code == GT ? GE : LT; return; } @@ -5069,7 +5069,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, if (i != ~((unsigned HOST_WIDE_INT) 0) && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) { - *op1 = GEN_INT (i + 1); + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); *code = *code == GTU ? GEU : LTU; return; } diff --git a/gcc/testsuite/g++.dg/torture/pr65694.C b/gcc/testsuite/g++.dg/torture/pr65694.C new file mode 100644 index 0000000..55e009a --- /dev/null +++ b/gcc/testsuite/g++.dg/torture/pr65694.C @@ -0,0 +1,144 @@ +/* { dg-do compile } */ +/* { dg-options "-Wno-sign-compare -Wno-return-type -Wno-overflow" } */ +/* { dg-additional-options "-mthumb" { target arm_thumb2_ok } } */ + +struct A { + enum { __value }; +}; +template <class _T1> struct B { _T1 first; }; +template <typename _Iterator, bool> struct C { + typedef typename _Iterator::iterator_type iterator_type; + static iterator_type _S_base(_Iterator p1) { return p1.base(); } +}; +template <typename _RandomAccessIterator> +typename _RandomAccessIterator::difference_type +__distance(_RandomAccessIterator p1, _RandomAccessIterator p2, int) { + return p2 - p1; +} + +template <typename _InputIterator> +typename _InputIterator::difference_type distance(_InputIterator p1, + _InputIterator p2) { + return __distance(p1, p2, 0); +} + +template <typename _Iterator, typename> class D { + _Iterator _M_current; + +public: + typedef _Iterator iterator_type; + typedef int difference_type; + _Iterator base() { return _M_current; } +}; + +template <typename _Iterator, typename _Container> +typename D<_Iterator, _Container>::difference_type +operator-(D<_Iterator, _Container> p1, D<_Iterator, _Container> p2) { + return p1.base() - p2.base(); +} + +struct F { + static unsigned short *__copy_m(unsigned short *p1, unsigned short *p2, + unsigned short *p3) { + int a = p2 - p1; + if (a) + __builtin_memmove(p3, p1, a); + return p3 + a; + } +}; +class G { +public: + void allocate(int p1) { + if (p1 > max_size()) + operator new(sizeof(short)); + } + unsigned max_size() { return -1 / sizeof(short); } +}; + +template <typename> class L : public G {}; + +struct H { + static unsigned short *allocate(int p1) { + L<short> d; + d.allocate(p1); + } +}; +struct I { + template <typename _InputIterator, typename _ForwardIterator> + static _ForwardIterator __uninit_copy(_InputIterator p1, _InputIterator p2, + _ForwardIterator p3) { + return copy(p1, p2, p3); + } +}; +struct J { + typedef unsigned short *pointer; + struct K { + unsigned short *_M_start; + unsigned short *_M_finish; + }; + J(); + J(int p1, int) { _M_create_storage(p1); } + K _M_impl; + pointer _M_allocate(unsigned p1) { p1 ? H::allocate(p1) : pointer(); } + void _M_create_storage(int p1) { _M_allocate(p1); } +}; + +C<D<unsigned short *, int>, 1>::iterator_type +__miter_base(D<unsigned short *, int> p1) { + return C<D<unsigned short *, int>, 1>::_S_base(p1); +} + +template <bool, typename _II, typename _OI> +_OI __copy_move_a(_II p1, _II p2, _OI p3) { + return F::__copy_m(p1, p2, p3); +} + +template <bool _IsMove, typename _II, typename _OI> +_OI __copy_move_a2(_II p1, _II p2, _OI p3) { + return __copy_move_a<_IsMove>(p1, p2, p3); +} + +template <typename _II, typename _OI> _OI copy(_II p1, _II p2, _OI p3) { + C<D<unsigned short *, int>, 1>::iterator_type b, c = __miter_base(p1); + b = __miter_base(p2); + return __copy_move_a2<A::__value>(c, b, p3); +} + +template <typename _InputIterator, typename _ForwardIterator> +_ForwardIterator uninitialized_copy(_InputIterator p1, _InputIterator p2, + _ForwardIterator p3) { + return I::__uninit_copy(p1, p2, p3); +} + +template <typename _InputIterator, typename _ForwardIterator, typename _Tp> +_ForwardIterator __uninitialized_copy_a(_InputIterator p1, _InputIterator p2, + _ForwardIterator p3, L<_Tp>) { + return uninitialized_copy(p1, p2, p3); +} + +class M : J { + J _Base; + +public: + M(); + M(int p1, int p2 = int()) : _Base(p1, p2) {} + M(D<unsigned short *, int> p1, D<unsigned short *, int> p2) { + _M_initialize_dispatch(p1, p2, int()); + } + D<pointer, int> begin(); + D<pointer, int> end(); + int size() { return _M_impl._M_finish - _M_impl._M_start; } + void _M_initialize_dispatch(D<unsigned short *, int> p1, + D<unsigned short *, int> p2, int) { + L<short> e; + int f = distance(p1, p2); + _M_impl._M_start = _M_allocate(f); + _M_impl._M_finish = __uninitialized_copy_a(p1, p2, _M_impl._M_start, e); + } +}; + +B<M> g, h; +void twoMeans() { + M i(g.first.begin(), h.first.end()); + M(i.size()); +} ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH][ARM] PR 65694: Properly sign-extend large numbers before passing to GEN_INT in arm_canonicalize_comparison 2015-04-10 8:33 [PATCH][ARM] PR 65694: Properly sign-extend large numbers before passing to GEN_INT in arm_canonicalize_comparison Kyrill Tkachov @ 2015-04-10 8:57 ` Jakub Jelinek 2015-04-10 11:06 ` Kyrill Tkachov 0 siblings, 1 reply; 4+ messages in thread From: Jakub Jelinek @ 2015-04-10 8:57 UTC (permalink / raw) To: Kyrill Tkachov; +Cc: GCC Patches, Ramana Radhakrishnan, Richard Earnshaw On Fri, Apr 10, 2015 at 09:33:11AM +0100, Kyrill Tkachov wrote: > 2015-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com> > Missing PR target/65694 line here. > * config/arm/arm.c (arm_canonicalize_comparison): Use ARM_SIGN_EXTEND > when creating +1 values for SImode and trunc_int_for_mode for similar > DImode operations. > > 2015-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com> > Ditto. > * g++.dg/torture/pr65694.C: New test. > diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c > index 369cb67..5342b33 100644 > --- a/gcc/config/arm/arm.c > +++ b/gcc/config/arm/arm.c > @@ -4984,7 +4984,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, > if (i != maxval > && arm_const_double_by_immediates (GEN_INT (i + 1))) > { > - *op1 = GEN_INT (i + 1); > + *op1 = GEN_INT (trunc_int_for_mode (i + 1, DImode)); > *code = *code == GT ? GE : LT; > return; > } > @@ -4994,7 +4994,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, > if (i != ~((unsigned HOST_WIDE_INT) 0) > && arm_const_double_by_immediates (GEN_INT (i + 1))) > { > - *op1 = GEN_INT (i + 1); > + *op1 = GEN_INT (trunc_int_for_mode (i + 1, DImode)); The above two aren't strictly necessary, HOST_WIDE_INT is always 64-bit, so is DImode, and GEN_INT takes HOST_WIDE_INT. You haven't changed it in the GEN_INT (i + 1) calls passed to arm_const_double_by_immediates anyway. I'd think you can leave those changes to cleanup in stage1 if desirable. > @@ -5047,7 +5047,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, > if (i != maxval > && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) > { > - *op1 = GEN_INT (i + 1); > + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); > *code = *code == GT ? GE : LT; > return; > } > @@ -5069,7 +5069,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, > if (i != ~((unsigned HOST_WIDE_INT) 0) > && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) > { > - *op1 = GEN_INT (i + 1); > + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); > *code = *code == GTU ? GEU : LTU; > return; > } This looks ok to me, but I'll defer the final word to ARM maintainers. That said, the ARM_SIGN_EXTEND macro could very well use some cleanup too now that HOST_WIDE_INT is always 64-bit and one can e.g. use HOST_WIDE_INT_{U,}C macros to build large constants. Jakub ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH][ARM] PR 65694: Properly sign-extend large numbers before passing to GEN_INT in arm_canonicalize_comparison 2015-04-10 8:57 ` Jakub Jelinek @ 2015-04-10 11:06 ` Kyrill Tkachov 2015-04-10 16:11 ` Jakub Jelinek 0 siblings, 1 reply; 4+ messages in thread From: Kyrill Tkachov @ 2015-04-10 11:06 UTC (permalink / raw) To: Jakub Jelinek; +Cc: GCC Patches, Ramana Radhakrishnan, Richard Earnshaw [-- Attachment #1: Type: text/plain, Size: 3337 bytes --] Hi Jakub, On 10/04/15 09:57, Jakub Jelinek wrote: > On Fri, Apr 10, 2015 at 09:33:11AM +0100, Kyrill Tkachov wrote: >> 2015-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com> >> > Missing > PR target/65694 > line here. Fixed. > >> * config/arm/arm.c (arm_canonicalize_comparison): Use ARM_SIGN_EXTEND >> when creating +1 values for SImode and trunc_int_for_mode for similar >> DImode operations. >> >> 2015-04-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com> >> > Ditto. Fixed. > >> * g++.dg/torture/pr65694.C: New test. >> diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c >> index 369cb67..5342b33 100644 >> --- a/gcc/config/arm/arm.c >> +++ b/gcc/config/arm/arm.c >> @@ -4984,7 +4984,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, >> if (i != maxval >> && arm_const_double_by_immediates (GEN_INT (i + 1))) >> { >> - *op1 = GEN_INT (i + 1); >> + *op1 = GEN_INT (trunc_int_for_mode (i + 1, DImode)); >> *code = *code == GT ? GE : LT; >> return; >> } >> @@ -4994,7 +4994,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, >> if (i != ~((unsigned HOST_WIDE_INT) 0) >> && arm_const_double_by_immediates (GEN_INT (i + 1))) >> { >> - *op1 = GEN_INT (i + 1); >> + *op1 = GEN_INT (trunc_int_for_mode (i + 1, DImode)); > The above two aren't strictly necessary, HOST_WIDE_INT is always 64-bit, so > is DImode, and GEN_INT takes HOST_WIDE_INT. Yeah, those aren't strictly necessary, it's the SImode code that was causing the ICE. > You haven't changed it in the GEN_INT (i + 1) calls passed to > arm_const_double_by_immediates anyway. > I'd think you can leave those changes to cleanup in stage1 if desirable. > >> @@ -5047,7 +5047,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, >> if (i != maxval >> && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) >> { >> - *op1 = GEN_INT (i + 1); >> + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); >> *code = *code == GT ? GE : LT; >> return; >> } >> @@ -5069,7 +5069,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, >> if (i != ~((unsigned HOST_WIDE_INT) 0) >> && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) >> { >> - *op1 = GEN_INT (i + 1); >> + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); >> *code = *code == GTU ? GEU : LTU; >> return; >> } > This looks ok to me, but I'll defer the final word to ARM maintainers. > That said, the ARM_SIGN_EXTEND macro could very well use some cleanup too > now that HOST_WIDE_INT is always 64-bit and one can e.g. use > HOST_WIDE_INT_{U,}C macros to build large constants. But I'd think that's stage 1 material anyway. Also, trunc_int_for_mode SImode would have worked here, I suspect. Attached is updated patch without the DImode stuff and ChangeLog. Thanks, Kyrill 2015-04-09 Kyrylo Tkachov<kyrylo.tkachov@arm.com> PR target/65694 * config/arm/arm.c (arm_canonicalize_comparison): Use ARM_SIGN_EXTEND when creating +1 values for SImode. 2015-04-09 Kyrylo Tkachov<kyrylo.tkachov@arm.com> PR target/65694 * g++.dg/torture/pr65694.C: New test. [-- Warning: decoded text below may be mangled, UTF-8 assumed --] [-- Attachment #2: arm-comp-signextend.patch --] [-- Type: text/x-patch; name=arm-comp-signextend.patch, Size: 5680 bytes --] commit 656485123eb7dbcdd03889603cf5ed7c43f7d553 Author: Kyrylo Tkachov <kyrylo.tkachov@arm.com> Date: Wed Apr 8 17:08:17 2015 +0100 [ARM] PR target/65694: Use ARM_SIGN_EXTEND in arm_canonicalize_comparison diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c index 10edeb4..77b9917 100644 --- a/gcc/config/arm/arm.c +++ b/gcc/config/arm/arm.c @@ -5047,7 +5047,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, if (i != maxval && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) { - *op1 = GEN_INT (i + 1); + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); *code = *code == GT ? GE : LT; return; } @@ -5069,7 +5069,7 @@ arm_canonicalize_comparison (int *code, rtx *op0, rtx *op1, if (i != ~((unsigned HOST_WIDE_INT) 0) && (const_ok_for_arm (i + 1) || const_ok_for_arm (-(i + 1)))) { - *op1 = GEN_INT (i + 1); + *op1 = GEN_INT (ARM_SIGN_EXTEND (i + 1)); *code = *code == GTU ? GEU : LTU; return; } diff --git a/gcc/testsuite/g++.dg/torture/pr65694.C b/gcc/testsuite/g++.dg/torture/pr65694.C new file mode 100644 index 0000000..55e009a --- /dev/null +++ b/gcc/testsuite/g++.dg/torture/pr65694.C @@ -0,0 +1,144 @@ +/* { dg-do compile } */ +/* { dg-options "-Wno-sign-compare -Wno-return-type -Wno-overflow" } */ +/* { dg-additional-options "-mthumb" { target arm_thumb2_ok } } */ + +struct A { + enum { __value }; +}; +template <class _T1> struct B { _T1 first; }; +template <typename _Iterator, bool> struct C { + typedef typename _Iterator::iterator_type iterator_type; + static iterator_type _S_base(_Iterator p1) { return p1.base(); } +}; +template <typename _RandomAccessIterator> +typename _RandomAccessIterator::difference_type +__distance(_RandomAccessIterator p1, _RandomAccessIterator p2, int) { + return p2 - p1; +} + +template <typename _InputIterator> +typename _InputIterator::difference_type distance(_InputIterator p1, + _InputIterator p2) { + return __distance(p1, p2, 0); +} + +template <typename _Iterator, typename> class D { + _Iterator _M_current; + +public: + typedef _Iterator iterator_type; + typedef int difference_type; + _Iterator base() { return _M_current; } +}; + +template <typename _Iterator, typename _Container> +typename D<_Iterator, _Container>::difference_type +operator-(D<_Iterator, _Container> p1, D<_Iterator, _Container> p2) { + return p1.base() - p2.base(); +} + +struct F { + static unsigned short *__copy_m(unsigned short *p1, unsigned short *p2, + unsigned short *p3) { + int a = p2 - p1; + if (a) + __builtin_memmove(p3, p1, a); + return p3 + a; + } +}; +class G { +public: + void allocate(int p1) { + if (p1 > max_size()) + operator new(sizeof(short)); + } + unsigned max_size() { return -1 / sizeof(short); } +}; + +template <typename> class L : public G {}; + +struct H { + static unsigned short *allocate(int p1) { + L<short> d; + d.allocate(p1); + } +}; +struct I { + template <typename _InputIterator, typename _ForwardIterator> + static _ForwardIterator __uninit_copy(_InputIterator p1, _InputIterator p2, + _ForwardIterator p3) { + return copy(p1, p2, p3); + } +}; +struct J { + typedef unsigned short *pointer; + struct K { + unsigned short *_M_start; + unsigned short *_M_finish; + }; + J(); + J(int p1, int) { _M_create_storage(p1); } + K _M_impl; + pointer _M_allocate(unsigned p1) { p1 ? H::allocate(p1) : pointer(); } + void _M_create_storage(int p1) { _M_allocate(p1); } +}; + +C<D<unsigned short *, int>, 1>::iterator_type +__miter_base(D<unsigned short *, int> p1) { + return C<D<unsigned short *, int>, 1>::_S_base(p1); +} + +template <bool, typename _II, typename _OI> +_OI __copy_move_a(_II p1, _II p2, _OI p3) { + return F::__copy_m(p1, p2, p3); +} + +template <bool _IsMove, typename _II, typename _OI> +_OI __copy_move_a2(_II p1, _II p2, _OI p3) { + return __copy_move_a<_IsMove>(p1, p2, p3); +} + +template <typename _II, typename _OI> _OI copy(_II p1, _II p2, _OI p3) { + C<D<unsigned short *, int>, 1>::iterator_type b, c = __miter_base(p1); + b = __miter_base(p2); + return __copy_move_a2<A::__value>(c, b, p3); +} + +template <typename _InputIterator, typename _ForwardIterator> +_ForwardIterator uninitialized_copy(_InputIterator p1, _InputIterator p2, + _ForwardIterator p3) { + return I::__uninit_copy(p1, p2, p3); +} + +template <typename _InputIterator, typename _ForwardIterator, typename _Tp> +_ForwardIterator __uninitialized_copy_a(_InputIterator p1, _InputIterator p2, + _ForwardIterator p3, L<_Tp>) { + return uninitialized_copy(p1, p2, p3); +} + +class M : J { + J _Base; + +public: + M(); + M(int p1, int p2 = int()) : _Base(p1, p2) {} + M(D<unsigned short *, int> p1, D<unsigned short *, int> p2) { + _M_initialize_dispatch(p1, p2, int()); + } + D<pointer, int> begin(); + D<pointer, int> end(); + int size() { return _M_impl._M_finish - _M_impl._M_start; } + void _M_initialize_dispatch(D<unsigned short *, int> p1, + D<unsigned short *, int> p2, int) { + L<short> e; + int f = distance(p1, p2); + _M_impl._M_start = _M_allocate(f); + _M_impl._M_finish = __uninitialized_copy_a(p1, p2, _M_impl._M_start, e); + } +}; + +B<M> g, h; +void twoMeans() { + M i(g.first.begin(), h.first.end()); + M(i.size()); +} ^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH][ARM] PR 65694: Properly sign-extend large numbers before passing to GEN_INT in arm_canonicalize_comparison 2015-04-10 11:06 ` Kyrill Tkachov @ 2015-04-10 16:11 ` Jakub Jelinek 0 siblings, 0 replies; 4+ messages in thread From: Jakub Jelinek @ 2015-04-10 16:11 UTC (permalink / raw) To: Kyrill Tkachov; +Cc: GCC Patches, Ramana Radhakrishnan, Richard Earnshaw On Fri, Apr 10, 2015 at 12:06:31PM +0100, Kyrill Tkachov wrote: > >This looks ok to me, but I'll defer the final word to ARM maintainers. > >That said, the ARM_SIGN_EXTEND macro could very well use some cleanup too > >now that HOST_WIDE_INT is always 64-bit and one can e.g. use > >HOST_WIDE_INT_{U,}C macros to build large constants. > > But I'd think that's stage 1 material anyway. Sure. > Attached is updated patch without the DImode stuff and ChangeLog. > > Thanks, > Kyrill > > 2015-04-09 Kyrylo Tkachov<kyrylo.tkachov@arm.com> > > PR target/65694 > * config/arm/arm.c (arm_canonicalize_comparison): Use ARM_SIGN_EXTEND > when creating +1 values for SImode. > > 2015-04-09 Kyrylo Tkachov<kyrylo.tkachov@arm.com> > > PR target/65694 > * g++.dg/torture/pr65694.C: New test. As no ARM maintainer popped up, I'll approve it for the trunk now. As PR65722 is still not resolved and PR65710 got reopened, I think I'm not going to do RC1 today, but would like to do it during the weekend or shorly afterwards. Jakub ^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2015-04-10 16:11 UTC | newest] Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2015-04-10 8:33 [PATCH][ARM] PR 65694: Properly sign-extend large numbers before passing to GEN_INT in arm_canonicalize_comparison Kyrill Tkachov 2015-04-10 8:57 ` Jakub Jelinek 2015-04-10 11:06 ` Kyrill Tkachov 2015-04-10 16:11 ` Jakub Jelinek
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).