[Bug target/101142] [11/12 regression] Regression due to supporting bitwise operators on AVX512 masks.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101142 --- Comment #4 from Hongtao.liu --- (In reply to Andrew Pinski from comment #3) > The exact command line to hit this issue is: > -O3 -march=skylake-avx512 Yes, thanks for the clarification. g++ byteswap.cpp test.cpp -march=skylake-avx512 -O3 -DDTYPE32
[Bug target/100866] PPC: Inefficient code for vec_revb(vector unsigned short) < P9
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100866 --- Comment #9 from Jens Seifert --- I know that if I would use vec_perm builtin as an end user, that you then need to fulfill to the LE specification, but you can always optimize the code as you like as long as it creates correct results afterwards. load constant xxlnor constant can always be transformed to load inverse constant.
[Bug target/101142] [11/12 regression] Regression due to supporting bitwise operators on AVX512 masks.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101142 Andrew Pinski changed: What|Removed |Added Target Milestone|--- |11.2 Keywords||missed-optimization Last reconfirmed||2021-06-21 Status|UNCONFIRMED |NEW Ever confirmed|0 |1 --- Comment #3 from Andrew Pinski --- The exact command line to hit this issue is: -O3 -march=skylake-avx512
[Bug debug/101141] Fedora glibc debuginfo .dwz contains a partial unit with needed debuginfo but which is not imported
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101141 --- Comment #3 from roc at ocallahan dot org --- Filed https://sourceware.org/bugzilla/show_bug.cgi?id=28000
[Bug target/101142] [11/12 regression] Regression due to supporting bitwise operators on AVX512 masks.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101142
--- Comment #2 from Hongtao.liu ---
I'm working on a patch which disparages slightly the mask register alternative
for bitwise operations(using "?k" in alternatives). It can prevent mask bitwise
instruction generation when the input is not allocate as mask registers.
Also when allocano cost of GENERAL_REGS is same as MASK_REGS, allocate
MASK_REGS first since it has already been disparaged. This is for testcase like
below where the input is allocated as mask registers, then mask bitwise
instructions should be used here.
#include
volatile __mmask8 foo;
void
foo_orb (__m512i a, __m512i b, __m512i c, __m512i d)
{
__mmask8 m1 = _mm512_cmp_epi64_mask (a, b, 2);
__mmask8 m2 = _mm512_cmp_epi64_mask (c, d, 4);
foo = m1 | m2;
}
vpcmpq $2, %zmm1, %zmm0, %k0
vpcmpq $4, %zmm3, %zmm2, %k1
korb%k1, %k0, %k2
kmovb %k2, foo(%rip)
ret
foo:
[Bug debug/101141] Fedora glibc debuginfo .dwz contains a partial unit with needed debuginfo but which is not imported
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101141 roc at ocallahan dot org changed: What|Removed |Added Resolution|--- |INVALID Status|WAITING |RESOLVED --- Comment #2 from roc at ocallahan dot org --- Good point. It looks correct in the generated .so so it's probably a DWZ bug. I'll repost it there. Sorry for the noise.
[Bug target/101142] [11/12 regression] Regression due to supporting bitwise operators on AVX512 masks.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101142 --- Comment #1 from Hongtao.liu --- Created attachment 51041 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51041=edit byteswap.cpp
[Bug target/101142] New: [11/12 regression] Regression due to supporting bitwise operators on AVX512 masks.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101142 Bug ID: 101142 Summary: [11/12 regression] Regression due to supporting bitwise operators on AVX512 masks. Product: gcc Version: 12.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: target Assignee: unassigned at gcc dot gnu.org Reporter: crazylht at gmail dot com Target Milestone: --- Host: x86_64-pc-linux-gnu Target: x86_64-*-* i?86-*-* Created attachment 51040 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=51040=edit test.cpp options to reproduce regression: g++ byteswap.cpp test.cpp -march=native -O3 -DDTYPE32 The regression is due to r11-2796-g388cb292a94f98a276548cd6ce01285cf36d17df which supports bitwise operator for avx512 masks. In byteswap.cpp there're a lot of bitwise operations, the register pressure is very high, so LRA allocate some bitwise operations to mask registers to avoid spills, the problem is mask bitwise instructions has 1/4 throught of those gpr versions which causes the regression.
[Bug debug/101141] Fedora glibc debuginfo .dwz contains a partial unit with needed debuginfo but which is not imported
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101141 Andrew Pinski changed: What|Removed |Added Ever confirmed|0 |1 Last reconfirmed||2021-06-21 Status|UNCONFIRMED |WAITING --- Comment #1 from Andrew Pinski --- This sounds like either a binutils issue or a https://sourceware.org/dwz/ issue or a Fedora issue. Can you rebuild glibc in Fedora and look at the resulting glibc.so file before the dwz and debug info gets stripped?
[Bug debug/101141] New: Fedora glibc debuginfo .dwz contains a partial unit with needed debuginfo but which is not imported
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101141 Bug ID: 101141 Summary: Fedora glibc debuginfo .dwz contains a partial unit with needed debuginfo but which is not imported Product: gcc Version: 11.1.1 Status: UNCONFIRMED Severity: normal Priority: P3 Component: debug Assignee: unassigned at gcc dot gnu.org Reporter: roc at ocallahan dot org Target Milestone: --- The Fedora 34 package glibc-debuginfo-2.33-16.fc34.x86_64 package has glibc symbols in /usr/lib/debug/.build-id/08/1490fc18239fa63189a53e526a68ee5d19c571.debug whose .gnu_debugaltlink is /usr/lib/debug/usr/.dwz/glibc-2.33-16.fc34.x86_64. /usr/lib/debug/.build-id/08/1490fc18239fa63189a53e526a68ee5d19c571.debug has a compilation unit that imports the partial unit at 0x1f63c from the .dwz file: UNIT: < 0><0x000c GOFF=0x000795fa> DW_TAG_compile_unit DW_AT_producer <.debug_str(sup)+0x0001200e> DW_AT_language DW_LANG_C11 DW_AT_name global-locale.c DW_AT_comp_dir /usr/src/debug/glibc-2.33-16.fc34.x86_64/locale DW_AT_stmt_list <.debug_line+0xce8a> < 1><0x001e GOFF=0x0007960c>DW_TAG_imported_unit DW_AT_import <.debug_info(sup)+0x0001f63c> That partial unit contains a variable declaration whose DW_AT_specification (0x1f0d2) lives in another partial compilation unit: UNIT: < 0><0x000c GOFF=0x0001f63c> DW_TAG_partial_unit DW_AT_stmt_list <.debug_line+0x> ... < 1><0x0038 GOFF=0x0001f668>DW_TAG_variable DW_AT_specification <.debug_info+0x0001f0d2> DW_AT_decl_file 0x013b /usr/src/debug/glibc-2.33-16.fc34.x86_64/locale/global-locale.c DW_AT_decl_line 0x0040 DW_AT_decl_column 0x0013 DW_AT_location len 0x000a: 0e9b: DW_OP_const8u 0 DW_OP_form_tls_address UNIT: < 0><0x000c GOFF=0x0001f0cd> DW_TAG_partial_unit DW_AT_stmt_list <.debug_line+0x> < 1><0x0011 GOFF=0x0001f0d2>DW_TAG_variable DW_AT_name __libc_tsd_LOCALE DW_AT_decl_file 0x0089 /usr/src/debug/glibc-2.33-16.fc34.x86_64/locale/localeinfo.h DW_AT_decl_line 0x00e1 DW_AT_decl_column 0x0001 DW_AT_type <.debug_info+0x00035fb3> DW_AT_external yes DW_AT_declaration yes However, the partial unit at 0x1f0cd is not imported anywhere in /usr/lib/debug/.build-id/08/1490fc18239fa63189a53e526a68ee5d19c571.debug as far as I can tell. I think the compilation unit at global-locale.c, at least, should be importing it.
[Bug target/100866] PPC: Inefficient code for vec_revb(vector unsigned short) < P9
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100866
--- Comment #8 from luoxhu at gcc dot gnu.org ---
(In reply to Jens Seifert from comment #7)
> Regarding vec_revb for vector unsigned int. I agree that
> revb:
> .LFB0:
> .cfi_startproc
> vspltish %v1,8
> vspltisw %v0,-16
> vrlh %v2,%v2,%v1
> vrlw %v2,%v2,%v0
> blr
>
> works. But in this case, I would prefer the vperm approach assuming that the
> loaded constant for the permute vector can be re-used multiple times.
> But please get rid of the xxlnor 32,32,32. That does not make sense after
> loading a constant. Change the constant that need to be loaded.
xxlnor is LE specific requirement(not existed if build with -mbig), we need to
turn the index {0,1,2,3} to {31, 30,29,28} for vperm usage, it is required
otherwise produces incorrect result:
6|0x1630 <+16>:lvx v0,0,r9
7+> 0x1634 <+20>:xxlnor vs32,vs32,vs32
8|0x1638 <+24>:vperm v2,v2,v2,v0
9|0x163c <+28>:blr
(gdb)
0x1634 in revb ()
2: /x $vs34.uint128 = 0x42345678323456782234567812345678
5: /x $vs32.uint128 = 0xc0d0e0f08090a0b0405060700010203
(gdb) si
0x1638 in revb ()
2: /x $vs34.uint128 = 0x42345678323456782234567812345678
5: /x $vs32.uint128 = 0xf3f2f1f0f7f6f5f4fbfaf9f8fffefdfc
(gdb) si
0x163c in revb ()
2: /x $vs34.uint128 = 0x78563442785634327856342278563412
5: /x $vs32.uint128 = 0xf3f2f1f0f7f6f5f4fbfaf9f8fffefdfc
Quoted from the ISA:
vperm VRT,VRA,VRB,VRC
vsrc.qword[0] ← VSR[VRA+32]
vsrc.qword[1] ← VSR[VRB+32]
do i = 0 to 15
index ← VSR[VRC+32].byte[i].bit[3:7]
VSR[VRT+32].byte[i] ← src.byte[index]
end
Let the source vector be the concatenation of the
contents of VSR[VRA+32] followed by the contents of
VSR[VRB+32].
For each integer value i from 0 to 15, do the following.
Let index be the value specified by bits 3:7 of byte
element i of VSR[VRC+32].
The contents of byte element index of src are
placed into byte element i of VSR[VRT+32].
RISC-V: Parsing custom extension that is version 0
When giving gcc a -march string with a custom extension of version 0 (for example pulpv0) then gcc will think assign in the default version of 2p0. In gcc/common/config/riscv/riscv-common.c the function riscv_subset_list::parsing_subset_version falls back to the default version (2p0) when parsing if the major and minor version are both zero (which is the case for the string "pulpv0"). This means both "pulpv0" and "pulpv2" will get assigned the version 2p0. Looks wrong to me. Robert
[Bug fortran/100971] ICE: Bad IO basetype (7)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100971 Jerry DeLisle changed: What|Removed |Added CC||jvdelisle2 at gmail dot com --- Comment #1 from Jerry DeLisle --- I can confirm this bug.
[Bug rtl-optimization/46235] inefficient bittest code generation
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46235 Roger Sayle changed: What|Removed |Added CC||roger at nextmovesoftware dot com Resolution|--- |FIXED Target Milestone|--- |12.0 Status|UNCONFIRMED |RESOLVED --- Comment #7 from Roger Sayle --- Fixed on mainline.
[Bug libstdc++/101136] msdosdjgpp toolchain cannot find std::wstring_view
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101136 --- Comment #2 from Jonathan Wakely --- This is because the _GLIBCXX_USE_WCHAR_T macro is not defined, because etc are not complete on the target, so we don't have e.g. wcslen and other wchar_t functions. However, the wchar_t type is always defined for C++ and so the template can be used with it. So not a bug, but the expected behaviour. See mails from me last year about enabling more of the library without needing , which would fix this.
[Bug c++/101140] New: [modules] no matching function for call to ‘operator new(sizetype, void*)’
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101140
Bug ID: 101140
Summary: [modules] no matching function for call to ‘operator
new(sizetype, void*)’
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: ensadc at mailnesia dot com
Target Milestone: ---
$ cat new.hpp
extern "C++" {
void* operator new(__SIZE_TYPE__, void* p);
}
$ cat foo.cpp
module;
#include "new.hpp"
export module foo;
export template
T* construct_at(T* p) {
return ::new((void*)p) T();
}
$ cat bar.cpp
export module bar;
import foo;
void f(int* p) {
construct_at(p);
}
$ g++ -std=c++20 -fmodules-ts -c foo.cpp
$ g++ -std=c++20 -fmodules-ts -c bar.cpp
In module foo, imported at bar.cpp:3:
foo.cpp: In instantiation of ‘T* construct_at@foo(T*) [with T = int]’:
bar.cpp:6:17: required from here
foo.cpp:7:12: error: no matching function for call to ‘operator new(sizetype,
void*)’
7 | return ::new((void*)p) T();
|^~~
: note: candidate: ‘void* operator new(long unsigned int)’
: note: candidate expects 1 argument, 2 provided
: note: candidate: ‘void* operator new(long unsigned int,
std::align_val_t)’
: note: no known conversion for argument 2 from ‘void*’ to
‘std::align_val_t’
bar.cpp:1:8: warning: not writing module ‘bar’ due to errors
1 | export module bar;
|^~
The error disappears if "new.hpp" is included or imported in `bar.cpp`.
I originally encountered this problem when using `std::construct_at` (defined
in``) in a module.
[Bug target/101132] [11/12 regression] [MIPS/MSA] internal compiler error: in do_store_flag, at expr.c:12541
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101132 --- Comment #4 from Xi Ruoyao --- (In reply to Xi Ruoyao from comment #3) > Another testcase (produced by cvise from mesa-21.1.3): Flag: -O3 -mmsa -fno-trapping-math
[Bug target/101132] [11/12 regression] [MIPS/MSA] internal compiler error: in do_store_flag, at expr.c:12541
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101132
--- Comment #3 from Xi Ruoyao ---
Another testcase (produced by cvise from mesa-21.1.3):
unsigned float3_to_rgb9e5_gc_0;
util_format_r9g9b9e5_float_pack_rgba_float_dst_row_bc_0;
util_format_r9g9b9e5_float_pack_rgba_float_dst_row() {
unsigned x;
char *dst = util_format_r9g9b9e5_float_pack_rgba_float_dst_row;
for (; x; x += 1) {
int __trans_tmp_1, maxrgb_0;
struct {
unsigned u
} f, max;
if (f.u > 80)
__trans_tmp_1 = 0;
else
__trans_tmp_1 = max.u;
maxrgb_0 = __trans_tmp_1 > float3_to_rgb9e5_gc_0
?: util_format_r9g9b9e5_float_pack_rgba_float_dst_row_bc_0;
*dst = maxrgb_0;
dst += 4;
}
}
[Bug target/101132] [11/12 regression] [MIPS/MSA] internal compiler error: in do_store_flag, at expr.c:12541
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101132 Xi Ruoyao changed: What|Removed |Added Component|middle-end |target --- Comment #2 from Xi Ruoyao --- This seems mips specific.
[Bug tree-optimization/66787] gcc fails tail call elimination
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66787 Andrew Pinski changed: What|Removed |Added Known to fail||8.3.0 Component|c++ |tree-optimization Keywords||missed-optimization Known to work||12.0 --- Comment #3 from Andrew Pinski --- Looks like it is fixed on the trunk: t9.cc.044t.tailr1:Eliminated tail recursion in bb 4 : _14 = Array::computeSubSize (this_10(D), _5); t9.cc.044t.tailr1:Eliminated tail recursion in bb 7 : BlendingTable::create (this_16(D), _11, 255, 255); t9.cc.044t.tailr1:Eliminated tail recursion in bb 5 : BlendingTable::create (this_16(D), dst_15, _10, 255); t9.cc.044t.tailr1:Eliminated tail recursion in bb 3 : BlendingTable::create (this_16(D), dst_15, src_13, _9); t9.cc.044t.tailr1:Eliminated tail recursion in bb 8 : BlendingTable::print (this_14(D), _10, 255, 255); t9.cc.044t.tailr1:Eliminated tail recursion in bb 6 : BlendingTable::print (this_14(D), dst_17, _9, 255); t9.cc.044t.tailr1:Eliminated tail recursion in bb 4 : BlendingTable::print (this_14(D), dst_17, src_16, _8); It was not working in GCC 8.0 though: t9.cc.043t.tailr1:Eliminated tail recursion in bb 4 : _15 = Array::computeSubSize (this_12, _5); t9.cc.043t.tailr1:Eliminated tail recursion in bb 10 : BlendingTable::create (this_16, _11, 255, 255); t9.cc.043t.tailr1:Eliminated tail recursion in bb 8 : BlendingTable::create (this_16, dst_15, _10, 255); t9.cc.043t.tailr1:Eliminated tail recursion in bb 6 : BlendingTable::create (this_16, dst_15, src_13, _9);
[Bug middle-end/62062] Missed optimization: write ptr reloaded in each loop iteration
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62062 Andrew Pinski changed: What|Removed |Added Component|inline-asm |middle-end --- Comment #4 from Andrew Pinski --- for write_run_char_ptr_ref we get: [local count: 955630225]: # n_18 = PHI _2 = *p_10(D); _3 = _2 + 1; *p_10(D) = _3; *_2 = _13; n_8 = n_18 + -1; if (n_8 != -1(OVF)) goto ; [89.00%] else goto ; [11.00%] So I think this is still correct based on aliasing rules ... That is the what char* points to can be writing to a reference/pointer type of (char*&). Now we could version this loop for aliasing but I don't know how much it would benifit in general.
[Bug middle-end/61621] Normal enum switch slower than test case.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61621 Andrew Pinski changed: What|Removed |Added Component|c++ |middle-end Keywords||missed-optimization --- Comment #3 from Andrew Pinski --- Note I think this might have been improved already. Also I think there is an issue here which is not mentioned, in the case of "instructions", there might be most used ones which can be pulled out of the case table. Also in the below test function could even be using a constant table and a load from that based on the instructions [i] (and then added to value).
[Bug tree-optimization/94956] Unable to remove impossible ffs() test for zero
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94956 --- Comment #7 from Steinar H. Gunderson --- To wrap this up, confirming that GCC 11 does well on my benchmark: BM_Chain2054529 iterations 18781 ns/iter GCC 10, asm bsfq BM_Chain2044584 iterations 22509 ns/iter GCC 10, ffsll() BM_Chain2049753 iterations 20216 ns/iter GCC 11, asm bsfq BM_Chain2053346 iterations 18816 ns/iter GCC 11, ffsll() BM_Chain2064926 iterations 15747 ns/iter Clang 12, asm bsfq BM_Chain2071208 iterations 14374 ns/iter Clang 12, ffsll() So basically for 11+, the ffsll() statement does better than the bsfq statement, whereas it used to do markedly worse. Clang does even better, but I can live with that. :-)
[Bug tree-optimization/101139] Unable to remove double byteswap in fast path
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101139 --- Comment #2 from Andrew Pinski --- (In reply to Andrew Pinski from comment #1) > if (b_13 < h.0_15) > goto ; [51.12%] > else > goto ; [48.88%] > >[local count: 548896825]: > _16 = (short unsigned int) f$ab_14; > _17 = (int) _16; > _18 = __builtin_bswap16 (_17); > goto ; [100.00%] > >[local count: 524845000]: > k_22 = i (); > c.1_23 = (short unsigned int) k_22; > _24 = (int) c.1_23; > _25 = __builtin_bswap16 (_24); > >[local count: 1073741824]: > # prephitmp_32 = PHI <_18(4), _25(5)> > > Basically the same issue as PR 13563. Once that issue is fixed we should get: if (b_13 < h.0_15) goto ; [51.12%] else goto ; [48.88%] [local count: 548896825]: _16 = (short unsigned int) f$ab_14; goto ; [100.00%] [local count: 524845000]: k_22 = i (); c.1_23 = (short unsigned int) k_22; [local count: 1073741824]: # c.1_23 = PHI <_16(4), c.1_23(5)> _24 = (int) c.1_23; prephitmp_32 = __builtin_bswap16 (_32); _2 = (int) prephitmp_32; _3 = __builtin_bswap16 (_2); _4 = (int) _3; Which then should just optimize later on to: if (b_13 < h.0_15) goto ; [51.12%] else goto ; [48.88%] [local count: 548896825]: _16 = (short unsigned int) f$ab_14; goto ; [100.00%] [local count: 524845000]: k_22 = i (); c.1_23 = (short unsigned int) k_22; [local count: 1073741824]: # c.1_23 = PHI <_16(4), c.1_23(5)> _24 = (int) c.1_23; _4 = _24; (If I did this conversion right)
[Bug tree-optimization/101139] Unable to remove double byteswap in fast path
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101139 Andrew Pinski changed: What|Removed |Added Last reconfirmed||2021-06-20 Assignee|unassigned at gcc dot gnu.org |pinskia at gcc dot gnu.org Depends on||13563 Status|UNCONFIRMED |ASSIGNED Keywords||missed-optimization Ever confirmed|0 |1 --- Comment #1 from Andrew Pinski --- if (b_13 < h.0_15) goto ; [51.12%] else goto ; [48.88%] [local count: 548896825]: _16 = (short unsigned int) f$ab_14; _17 = (int) _16; _18 = __builtin_bswap16 (_17); goto ; [100.00%] [local count: 524845000]: k_22 = i (); c.1_23 = (short unsigned int) k_22; _24 = (int) c.1_23; _25 = __builtin_bswap16 (_24); [local count: 1073741824]: # prephitmp_32 = PHI <_18(4), _25(5)> Basically the same issue as PR 13563. Referenced Bugs: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=13563 [Bug 13563] if-conversion not agressive enough
[Bug c++/61245] #pragma GCC ivdep is ignored with call inside the test of a for loop
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61245 --- Comment #5 from Andrew Pinski --- The loop is still vectorized though. As it looks like it was versioned. t6.cc: In function ‘void doT(SoA&) [with int N = 3]’: t6.cc:34:17: warning: ignoring loop annotation 34 | for (auto i=0U; i
[Bug tree-optimization/101139] New: Unable to remove double byteswap in fast path
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101139
Bug ID: 101139
Summary: Unable to remove double byteswap in fast path
Product: gcc
Version: 10.2.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: steinar+gcc at gunderson dot no
Target Milestone: ---
The following code is reduced from a real interpreter:
extern void (*a[])();
int d, e, h, l;
typedef struct {
char ab;
} f;
f g;
short i();
short m68ki_read_imm_16() {
short j, k;
int b = d;
f f = g;
if (b < h)
return __builtin_bswap16(()[0]);
k = i();
short c = k;
j = __builtin_bswap16(c);
return j;
}
int b() {
short m;
do {
m = m68ki_read_imm_16();
short c = m;
l = __builtin_bswap16(c);
a[l]();
} while (e);
return e;
}
Compiling with arm-linux-gnueabihf-gcc-10 -O2 yields this interesting sequence
in the function:
b .L11
.L15:
ldrbr3, [r5, #8]@ zero_extendqisi2
rev16 r3, r3
uxthr3, r3
.L10:
rev16 r3, r3
uxthr3, r3
The original code intention was to have a reusable function that returned in
big-endian, but that a specific use of it would be able to ignore endianness
into a table lookup, removing the double-swap entirely. GCC can normally do
that, but it seems that the branch in m68ki_read_imm_16() somehow gets in the
way. Just to be clear, I expect zero rev16 instructions altogether in b() when
m68ki_read_imm_16() is inlined.
The problem is not ARM-specific; x86 shows a similar problematic sequence:
leaqa(%rip), %rbx
jmp .L11
.p2align 4,,10
.p2align 3
.L15:
movsbw g(%rip), %ax
rolw$8, %ax
.L10:
rolw$8, %ax
movzwl %ax, %edx
Also verified with
gcc version 12.0.0 20210527 (experimental) [master revision
262e75d22c3:7bb6b9b2f47:9d3a953ec4d2695e9a6bfa5f22655e2aea47a973] (Debian
20210527-1)
[Bug c++/61245] #pragma GCC ivdep is ignored with call inside the test of a for loop
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61245 Andrew Pinski changed: What|Removed |Added Summary|ICE at in expand_ANNOTATE, |#pragma GCC ivdep is |at internal-fn.c:127 called |ignored with call inside |from cfgexpand.c|the test of a for loop --- Comment #4 from Andrew Pinski --- The ICE to warning was fixed with r5-4959.
[Bug c++/61245] ICE at in expand_ANNOTATE, at internal-fn.c:127 called from cfgexpand.c
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61245 Andrew Pinski changed: What|Removed |Added Keywords|ice-on-valid-code |missed-optimization --- Comment #3 from Andrew Pinski --- Do we get a warning now instead of an internal compiler error: t6.cc:34:17: warning: ignoring loop annotation 34 | for (auto i=0U; i
[Bug target/59555] bogus error: template with C linkage with preprocessed c++ file
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59555 Andrew Pinski changed: What|Removed |Added Resolution|--- |FIXED Component|c++ |target Status|UNCONFIRMED |RESOLVED Target Milestone|--- |9.0 --- Comment #1 from Andrew Pinski --- Fixed for GCC 9 by r9-1648 which changed NO_IMPLICIT_EXTERN_C to SYSTEM_IMPLICIT_EXTERN_C and made it less fragile.
[Bug target/56066] g++ generates strong symbols conflicting with C99 extern inline code on Windows
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56066 Andrew Pinski changed: What|Removed |Added Resolution|--- |INVALID Status|REOPENED|RESOLVED Component|c++ |target --- Comment #6 from Andrew Pinski --- I don't think this is a bug. Rather the problem is you are linking two different linkages and expecting it to work. in the C++ case, it is vague linkage while in C, there is extern linkage still. the correct thing is to use gnu_inline so it is the linkage you expect in both langauges.
[Bug c++/43064] improve location and text of diagnostics in constructor initializer lists
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=43064 Andrew Pinski changed: What|Removed |Added CC||bero at arklinux dot org --- Comment #13 from Andrew Pinski --- *** Bug 43933 has been marked as a duplicate of this bug. ***
[Bug c++/43933] Suboptimal error message when supplying a bad default value in initialization
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=43933 Andrew Pinski changed: What|Removed |Added Resolution|--- |DUPLICATE Status|NEW |RESOLVED --- Comment #4 from Andrew Pinski --- In the end this is a dup of bug 43064 which is fixed in GCC 9 as I shown. *** This bug has been marked as a duplicate of bug 43064 ***
[Bug c++/43933] Suboptimal error message when supplying a bad default value in initialization
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=43933
Andrew Pinski changed:
What|Removed |Added
Known to fail||8.3.1
--- Comment #3 from Andrew Pinski ---
Looks like it has been fixed on the trunk:
t7.cc: In constructor ‘A::A()’:
t7.cc:13:10: error: call of overloaded ‘QString(int)’ is ambiguous
13 | A::A() : a(0), b(0) { }
| ^~~~
t7.cc:4:3: note: candidate: ‘QString::QString(char)’
4 | QString(char);
| ^~~
t7.cc:3:3: note: candidate: ‘QString::QString(const QString&)’
3 | QString(const QString&);
| ^~~
t7.cc:2:3: note: candidate: ‘QString::QString(const char*)’
2 | QString(const char*);
| ^~~
t7.cc:13:16: error: call of overloaded ‘QString(int)’ is ambiguous
13 | A::A() : a(0), b(0) { }
|^~~~
t7.cc:4:3: note: candidate: ‘QString::QString(char)’
4 | QString(char);
| ^~~
t7.cc:3:3: note: candidate: ‘QString::QString(const QString&)’
3 | QString(const QString&);
| ^~~
t7.cc:2:3: note: candidate: ‘QString::QString(const char*)’
2 | QString(const char*);
| ^~~
It was broken in GCC 8.3 though:
t7.cc: In constructor ‘A::A()’:
t7.cc:13:19: error: call of overloaded ‘QString(int)’ is ambiguous
A::A() : a(0), b(0) { }
^
[Bug c++/43881] warning attached to a function is emitted even though the function is not being called
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=43881
--- Comment #6 from Andrew Pinski ---
The best way to do this is to use asm instead:
extern "C" int close(int);
extern __typeof__ (close) close __attribute__ ((__warning__ ("The symbol close
refers to the system function. Use safe_close instead.")));
extern __typeof__ (close) safe_close asm("close");
[Bug c++/100134] [modules] ICE when using a vector in a module
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100134
--- Comment #1 from ensadc at mailnesia dot com ---
Reduced:
vector
namespace std {
template struct __replace_first_arg;
template class _Template, typename _Up,
typename _Tp, typename... _Types>
struct __replace_first_arg<_Template<_Tp, _Types...>, _Up> {
using type = _Template<_Up, _Types...>;
};
template struct allocator {
typedef _Tp value_type;
friend constexpr bool operator==(const allocator &,
const allocator &) noexcept {
return true;
}
};
} // namespace std
namespace __gnu_cxx {
template
struct __alloc_traits {
template struct rebind {
typedef std::__replace_first_arg<_Alloc, _Tp> other;
};
};
} // namespace __gnu_cxx
namespace std {
template > class vector {
typedef
typename __gnu_cxx::__alloc_traits<_Alloc>::template rebind<_Tp>::other
_Tp_alloc_type;
};
} // namespace std
foo.cpp
export module foo;
import ;
export struct Foo {
std::vector v;
};
build commands:
g++ -I . -x c++-system-header -fmodules-ts -std=c++20 vector
g++ -I . -std=c++20 -fmodules-ts foo.cpp
[Bug c++/43149] Partial optimization
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=43149 Andrew Pinski changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution|--- |INVALID --- Comment #9 from Andrew Pinski --- https://gcc.gnu.org/onlinedocs/gcc/Function-Specific-Option-Pragmas.html Works across headers as there is nothing special about headers really.
[Bug c++/67252] Demangler fails on template conversion operator
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67252 Andrew Pinski changed: What|Removed |Added Depends on||41233 --- Comment #2 from Andrew Pinski --- Might be a dup of bug 41233. Referenced Bugs: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=41233 [Bug 41233] Templated conversion operator produces symbol name that won't demangle
[Bug c++/101138] Ambiguous code (with operator==) compiled without error
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101138
--- Comment #1 from Andrew Pinski ---
ICC also accepts it but Microsoft rejects it.
But ICC rejects the following which GCC still accepts:
//#include
//using namespace std;
#define printf __builtin_printf
//template
struct D {
template bool operator==(Y a) const {
//cout << "f" <
bool operator==(T a, D b) {
printf("f2\n");
return false;
}
struct ok;
int main()
{
D a, b;
if (a == b)
return 0;
return 1;
}
[Bug c++/101138] New: Ambiguous code (with operator==) compiled without error
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101138
Bug ID: 101138
Summary: Ambiguous code (with operator==) compiled without
error
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: hiraditya at msn dot com
Target Milestone: ---
$ cat test.cpp
#include
using namespace std;
template
struct D {
template bool operator==(Y a) const {
cout << "f" <
bool operator==(T a, D b) {
cout << "fD" < a, b;
if (a == b)
return 0;
return 1;
}
gcc compiles this code fine, bug clang errors out.
https://godbolt.org/z/c13EExxeY
