[Bug ada/118712] [13/14/15 regression] segfault on missing body for primitive operator of derived real type
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118712 --- Comment #2 from Andrew Teylu --- As a note, the code is minimised into "being invalid". Here is a version that compiles without errors with 12.1 but still gives an ICE on 14.2.1: `stub.ads`: ``` package Stub is type GenT is delta 1.0 range 1.0 .. 10.0; function "-" (X : GenT; Y : GenT) return GenT; type DerT is new GenT; end Stub; ``` `stub.adb`: ``` package body Stub is function "-" (X : GenT; Y : GenT) return GenT is begin return GenT (X) - GenT (Y); end "-"; procedure Proc is Var : DerT; Initialised : Boolean := False; begin loop if (Initialised) then Var := 1.0 - Var; end if; Initialised := True; Var := 2.0; end loop; end Proc; end Stub; ``` Output with 12.1: ``` stub.adb:12:26: warning: "Var" may be referenced before it has a value [enabled by default] ``` With 14.2.1: ``` +===GNAT BUG DETECTED==+ | 14.2.1 20241007 [revision 4af44f2cf7d281f3e4f3957efce10e8b2ccb2ad3] (x86_64-suse-linux) | | Storage_Error stack overflow or erroneous memory access | | Error detected at stub.adb:6:4 | | Compiling stub.adb | | Please submit a bug report; see https://gcc.gnu.org/bugs/ . | | Use a subject line meaningful to you and us to track the bug.| | Include the entire contents of this bug box in the report. | | Include the exact command that you entered. | | Also include sources listed below. | +==+ Please include these source files with error report Note that list may not be accurate in some cases, so please double check that the problem can still be reproduced with the set of files listed. Consider also -gnatd.n switch (see debug.adb). stub.adb stub.ads compilation abandoned ```
[Bug ada/118712] New: "Storage_Error stack overflow or erroneous memory access" with ranged types
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118712 Bug ID: 118712 Summary: "Storage_Error stack overflow or erroneous memory access" with ranged types Product: gcc Version: 14.2.1 Status: UNCONFIRMED Severity: normal Priority: P3 Component: ada Assignee: unassigned at gcc dot gnu.org Reporter: andrew.teylu at vector dot com CC: dkm at gcc dot gnu.org Target Milestone: --- For the following Ada package: `stub.ads`: ``` package Stub is type GenT is delta 1.0 range 1.0 .. 10.0; function "-" (X : GenT; Y : GenT) return GenT; type DerT is new GenT; end Stub; ``` `stub.adb`: ``` package body Stub is procedure Proc is Var : DerT; begin Var := 1.0 - Var; end Proc; end Stub; ``` Compiling with gcc 14.2.1 gives: ``` +===GNAT BUG DETECTED==+ | 14.2.1 20241007 [revision 4af44f2cf7d281f3e4f3957efce10e8b2ccb2ad3] (x86_64-suse-linux) | | Storage_Error stack overflow or erroneous memory access | | Error detected at stub.adb:2:4 | | Compiling stub.adb | | Please submit a bug report; see https://gcc.gnu.org/bugs/ . | | Use a subject line meaningful to you and us to track the bug.| | Include the entire contents of this bug box in the report. | | Include the exact command that you entered. | | Also include sources listed below. | +==+ Please include these source files with error report Note that list may not be accurate in some cases, so please double check that the problem can still be reproduced with the set of files listed. Consider also -gnatd.n switch (see debug.adb). stub.adb stub.ads compilation abandoned ``` The same bug reproduces with 14.2.0 from Alire. I tested with a build of 12.1.0, and the bug does not reproduce there: ``` gcc --version gcc (GCC) 12.1.0 Copyright (C) 2022 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. ``` ``` stub.adb:1:14: error: missing body for "-" declared at stub.ads:3 stub.adb:5:20: warning: "Var" may be referenced before it has a value [enabled by default ```
[Bug debug/121157] -gcodeview does not work with Ada
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121157 --- Comment #10 from Andrew Teylu --- Btw, it isn't that it _never_ works: ``` package LOCAL_IO is function GET_NEXT return Integer; end LOCAL_IO; package body LOCAL_IO is type TEXT_TYPE (LENGTH : Positive := 1) is record null; end record; DESCRIPTOR : TEXT_TYPE; function GET_NEXT return Integer is begin return 1; end GET_NEXT; end LOCAL_IO; with LOCAL_IO; procedure Main is Value : Integer; begin null; Value := LOCAL_IO.GET_NEXT; end Main; ``` Steps: ``` gnatchop.exe -w input.txt gcc.exe -g -gcodeview -c *.adb gnatbind.exe main.ali gcc.exe -g -c b~main.adb gcc.exe -g -gcodeview -Wl,--pdb=main.pdb -g -O0 b~main.o main.o local_io.o -o main -L/home/avj/gnat-x86_64-windows64-15.1.0-2/lib/gcc/x86_64-w64-mingw32/15.1.0/adalib/ -lgnat -Wl,--subsystem,console ``` This does complete without error and generate `main.pdb`. It does crash if you use `-gcodeview` on the binder file, so I've omitted that.
[Bug debug/121157] -gcodeview does not work with Ada
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121157 --- Comment #11 from Andrew Teylu --- Ah, no, I think Eric is correct in that it doesn't work at all (even if gcc doesn't fail). I used https://github.com/Microsoft/microsoft-pdb/blob/master/cvdump/cvdump.exe to look at the pdb file generated when compiling/linking my example, and the `*** LINES` section looked like this: ``` ** Module: "D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/b~main.o" Mod::GetEnumLines failed ** Module: "D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/main.o" ** Module: "D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/local_io.o" ``` Notice: for the two files we compile with `-gcodeview` they _do not_ have `Mod::GetEnumLines failed` (which, I think, means there is PDB info for those objects, but it is empty). I then wrote a matching C version (including a binder main that I did not compile with `-gcodeview` to match what I did for Ada), and the `*** LINES` section looked like this: ``` ** Module: "D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/c/b_main.o" Mod::GetEnumLines failed ** Module: "D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/c/main.o" D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/c/main.c (MD5: 7935177CE228764361F5487B3BDA29AC), 0001:05E0-05F7, line/addr pairs = 3 4 05E0 5 05E8 6 05F0 ** Module: "D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/c/local_io.o" D:/DEV/msys64/home/ateylu/alire/tom_121157/v3/c/local_io.c (MD5: 5997CD3181362A6DA3EF9CCAA4E0B36D), 0001:0600-060B, line/addr pairs = 3 4 0600 5 0604 6 0609 ``` That is, there is the debug information for the object files (and, as we'd like, not for the binder main).
[Bug debug/121157] internal error on Ada's unconstrained array types with -gcodeview
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121157 --- Comment #7 from Andrew Teylu --- Yeah, so my reduced file no longer crashes if you use `-fgnat-encodings=all`. Re-reducing Tom's original files while `-fgnat-encodings=all` still gives a crash: ``` package LOCAL_IO is function GET_NEXT return Integer; end LOCAL_IO; package body LOCAL_IO is type TEXT_TYPE (LENGTH : Positive := 2) is record null; end record; DESCRIPTORS : array (1 .. 0) of TEXT_TYPE; function GET_NEXT return Integer is begin return 1; end GET_NEXT; end LOCAL_IO; ``` Steps: ``` gnatchop -w input.txt gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -c local_io.adb # no error gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -gcodeview -fgnat-encodings=all -c local_io.adb ``` Where the last gives: ``` +===GNAT BUG DETECTED==+ | 15.1.0 (x86_64-w64-mingw32) in AT_unsigned, at dwarf2out.cc:4572 | | Error detected around local_io.adb:10:1 | | Compiling local_io.adb | | Please submit a bug report; see https://gcc.gnu.org/bugs/ . | | Use a subject line meaningful to you and us to track the bug.| | Include the entire contents of this bug box in the report. | | Include the exact command that you entered. | | Also include sources listed below. | +==+ Please include these source files with error report Note that list may not be accurate in some cases, so please double check that the problem can still be reproduced with the set of files listed. Consider also -gnatd.n switch (see debug.adb). local_io.adb local_io.ads raised TYPES.UNRECOVERABLE_ERROR : comperr.adb:422 ```
[Bug ada/121162] New: GNAT hang with `-gcodeview -fgnat-encodings=all` (but not just with `-gcodeview`)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121162 Bug ID: 121162 Summary: GNAT hang with `-gcodeview -fgnat-encodings=all` (but not just with `-gcodeview`) Product: gcc Version: 15.1.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: ada Assignee: unassigned at gcc dot gnu.org Reporter: andrew.teylu at vector dot com CC: dkm at gcc dot gnu.org Target Milestone: --- For the following file: ``` package MINIMAL_HANG is type RANGE_TYPE is range 1 .. 11; function SOME_FUNCTION (str : String) return String; end MINIMAL_HANG; package body MINIMAL_HANG is type NULL_RECORD is record null; end record; ARRAY_VAR : array (RANGE_TYPE) of NULL_RECORD; function SOME_FUNCTION (str : String) return String is begin return str; end SOME_FUNCTION; end MINIMAL_HANG; ``` and with the following steps: ``` gnatchop -w input.txt gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -gcodeview -c minimal_hang.adb # doesn't hang gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -gcodeview -fgnat-encodings=all -c minimal_hang.adb # hangs ``` Then `gcc` doesn't appear to terminate. GCC version: ``` Using built-in specs. COLLECT_GCC=Z:\home\avj\gnat-x86_64-windows64-15.1.0-2\bin\gcc.exe COLLECT_LTO_WRAPPER=Z:/home/avj/gnat-x86_64-windows64-15.1.0-2/bin/../libexec/gcc/x86_64-w64- mingw32/15.1.0/lto-wrapper.exe Target: x86_64-w64-mingw32 Configured with: ../src/configure --prefix=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gcc/inst all --with-build-time-tools=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/binutils/install/bin -- enable-languages=c,ada,c++ --enable-libstdcxx --enable-libstdcxx-threads --enable-libada --di sable-nls --without-libiconv-prefix --disable-libstdcxx-pch --enable-lto --disable-multilib - -enable-threads=win32 --with-native-system-header-dir=/aaa/GNAT-FSF-builds/sbx/x86_64-windows 64/gcc/install/include --with-gnu-ld --with-gnu-as --with-mpfr=/aaa/GNAT-FSF-builds/sbx/x86_6 4-windows64/mpfr/install --with-gmp=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gmp/install --w ith-mpc=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/mpc/install --with-isl=/aaa/GNAT-FSF-builds /sbx/x86_64-windows64/isl/install --build=x86_64-w64-mingw32 Thread model: win32 Supported LTO compression algorithms: zlib gcc version 15.1.0 (GCC) ``` Compiler taken from here: https://github.com/alire-project/GNAT-FSF-builds/releases/tag/gnat-15.1.0-2
[Bug ada/121163] New: GNAT hang with `-g -gcodeview` (but not with just `-g`)
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121163 Bug ID: 121163 Summary: GNAT hang with `-g -gcodeview` (but not with just `-g`) Product: gcc Version: 15.1.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: ada Assignee: unassigned at gcc dot gnu.org Reporter: andrew.teylu at vector dot com CC: dkm at gcc dot gnu.org Target Milestone: --- For the following file: ``` package LOCAL_IO is function GET_NEXT return Integer; end LOCAL_IO; package body LOCAL_IO is type TEXT_TYPE is record null; end record; DESCRIPTORS : array (1 .. 2) of TEXT_TYPE; function GET_NEXT return Integer is begin return 1; end GET_NEXT; end LOCAL_IO; ``` and the following steps: ``` gnatchop -w input.txt gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -c local_io.adb # doesn't hang gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -gcodeview -c local_io.adb # hangs ``` Then `gcc` doesn't appear to terminate. GCC version: ``` Using built-in specs. COLLECT_GCC=Z:\home\avj\gnat-x86_64-windows64-15.1.0-2\bin\gcc.exe COLLECT_LTO_WRAPPER=Z:/home/avj/gnat-x86_64-windows64-15.1.0-2/bin/../libexec/gcc/x86_64-w64-mingw32/15.1.0/lto-wrapper.exe Target: x86_64-w64-mingw32 Configured with: ../src/configure --prefix=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gcc/install --with-build-time-tools=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/binutils/install/bin --enab le-languages=c,ada,c++ --enable-libstdcxx --enable-libstdcxx-threads --enable-libada --disable-nls --without-libiconv-prefix --disable-libstdcxx-pch --enable-lto --disable-multilib --enable- threads=win32 --with-native-system-header-dir=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gcc/install/include --with-gnu-ld --with-gnu-as --with-mpfr=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/ mpfr/install --with-gmp=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gmp/install --with-mpc=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/mpc/install --with-isl=/aaa/GNAT-FSF-builds/sbx/x86_64-wind ows64/isl/install --build=x86_64-w64-mingw32 Thread model: win32 Supported LTO compression algorithms: zlib gcc version 15.1.0 (GCC) ``` Compiler taken from here: https://github.com/alire-project/GNAT-FSF-builds/releases/tag/gnat-15.1.0-2
[Bug ada/121157] GNAT internal error when using -gcodeview
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121157 --- Comment #4 from Andrew Teylu --- As a note, my version of GCC comes from https://github.com/alire-project/GNAT-FSF-builds/releases/tag/gnat-15.1.0-2 ``` Using built-in specs. COLLECT_GCC=Z:\home\avj\gnat-x86_64-windows64-15.1.0-2\bin\gcc.exe COLLECT_LTO_WRAPPER=Z:/home/avj/gnat-x86_64-windows64-15.1.0-2/bin/../libexec/gcc/x86_64-w64-mingw32/15.1.0/lto-wrapper.exe Target: x86_64-w64-mingw32 Configured with: ../src/configure --prefix=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gcc/install --with-build-time-tools=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/binutils/install/bin --enab le-languages=c,ada,c++ --enable-libstdcxx --enable-libstdcxx-threads --enable-libada --disable-nls --without-libiconv-prefix --disable-libstdcxx-pch --enable-lto --disable-multilib --enable- threads=win32 --with-native-system-header-dir=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gcc/install/include --with-gnu-ld --with-gnu-as --with-mpfr=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/ mpfr/install --with-gmp=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/gmp/install --with-mpc=/aaa/GNAT-FSF-builds/sbx/x86_64-windows64/mpc/install --with-isl=/aaa/GNAT-FSF-builds/sbx/x86_64-wind ows64/isl/install --build=x86_64-w64-mingw32 Thread model: win32 Supported LTO compression algorithms: zlib gcc version 15.1.0 (GCC) ```
[Bug ada/121157] GNAT internal error when using -gcodeview
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121157 Andrew Teylu changed: What|Removed |Added CC||andrew.teylu at vector dot com --- Comment #3 from Andrew Teylu --- I reduced Tom's example to: ``` package LOCAL_IO is function GET_NEXT (str : String) return String; end LOCAL_IO; package body LOCAL_IO is function GET_NEXT (str : String) return String is begin return str; end GET_NEXT; end LOCAL_IO; ``` To check: ``` gnatchop -w input.txt gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -c local_io.adb # no error gnat-x86_64-windows64-15.1.0-2/bin/gcc.exe -g -gcodeview -c local_io.adb ``` Where the last gives: ``` +===GNAT BUG DETECTED==+ | 15.1.0 (x86_64-w64-mingw32) in AT_unsigned, at dwarf2out.cc:4572 | | Error detected around local_io.adb:6:1 | | Compiling local_io.adb | | Please submit a bug report; see https://gcc.gnu.org/bugs/ . | | Use a subject line meaningful to you and us to track the bug.| | Include the entire contents of this bug box in the report. | | Include the exact command that you entered. | | Also include sources listed below. | +==+ Please include these source files with error report Note that list may not be accurate in some cases, so please double check that the problem can still be reproduced with the set of files listed. Consider also -gnatd.n switch (see debug.adb). local_io.adb local_io.ads raised TYPES.UNRECOVERABLE_ERROR : comperr.adb:422 ``` What's interesting is `local_io.adb:6:1` is the start of the last line in the file (the error is always at the last line, irrespective of what's in the file).
