the rustc testsuite fails with 13.1 on i686-linux-gnu and powerpc64le-linux-gnu, apparently worked before with trunk 20230114, both reported on Debian and Ubuntu: https://bugs.launchpad.net/ubuntu/+source/rustc/+bug/2007642/comments/10 claims: /build/gdb-i4hVub/gdb-13.1/gdb/frame.c:2457: internal-error: inside_main_func: Assertion `block != nullptr' failed. A problem internal to GDB has been detected, further debugging may prove unreliable. https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1031745 same internal error
Created attachment 14714 [details] test case with readme Attaching code that reproduces the issue. See README file. Original source: https://salsa.debian.org/fg/rustc-gdb-1031745
With attached testcase I am able to reproduce the issue using gdb 13.1, but it works using 13.0.91.20230210. It might be related to the patch that fixed PR/30116.
Ugh. I'll investigate.
I confirm that reverting patch for PR/30116 makes the rust testsuite work again (for me).
The crash is in inside_main_function, due to the modifications here: commit 9370fd51ebfca8a8acacaecb92c57ee54f4f8382 gdb: detect main function even when there's no matching msymbol First, gdb doesn't find the minsym, probably because the msyms contain: [99] t 0x90c0 _ZN17rustc_gdb_10317454main17h5b5be7fe16a97225E section .text rustc_gdb_1031745::main::h5b5be7fe16a97225 zko06yobckx336v ... but we're looking for: (top-gdb) p name $13 = 0x292e0c0 "rustc_gdb_1031745::main" The _Z overloading strikes again. Anyway somehow we find a full symbol for this, but it doesn't have a block, so the assert fails.
rustc decides to emit both: <2><2f5b>: Abbrev Number: 10 (DW_TAG_subprogram) ... <2f6e> DW_AT_name : (indirect string, offset: 0x23a): main and <2><2f95>: Abbrev Number: 2 (DW_TAG_namespace) <2f96> DW_AT_name : (indirect string, offset: 0x23a): main in the same scope. Then the lookup finds the wrong one, because gdb's symbol tables are bad and still follow C namespace rules.
We can avoid the crash pretty easily but stopping the 'bt' before _start is harder.
Another thought here is it would probably be better on the whole to use the main symbol's linkage name; but that too seems like a larger change.
https://sourceware.org/pipermail/gdb-patches/2023-February/197370.html I'll apply to the gdb 13 branch as well when it goes in.
(In reply to Tom Tromey from comment #9) > https://sourceware.org/pipermail/gdb-patches/2023-February/197370.html Thanks! That worked with the test case.
Hello Tom, The testcase at https://salsa.debian.org/fg/rustc-gdb-1031745 has been update to add one more failing test case scenario described at https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1031946 Fabian reports: I updated the reproducer repo[0], broken and good output look like this: ----8<---- broken Breakpoint 1 at 0x81dd: file src/main.rs, line 88. [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Breakpoint 1, rustc_gdb_1031745::main () at src/main.rs:88 88 zzz(); // #break $1 = &rustc_gdb_1031745::Foo<[u8]> 0x7fffffffdd88 $2 = &rustc_gdb_1031745::Foo<rustc_gdb_1031745::Foo<[u8]>> 0x7fffffffdd88 $3 = &rustc_gdb_1031745::Foo<dyn core::fmt::Debug> {pointer: 0x555555593034, vtable: 0x5555555a3000warning: (Internal error: pc 0x5555555a3000 in read in CU, but not in symtab.) warning: (Error: pc 0x5555555a3000 in address map, but not in symtab.) } $4 = alloc::boxed::Box<rustc_gdb_1031745::Foo<dyn core::fmt::Debug>, alloc::alloc::Global> {pointer: 0x5555555a7ba0, vtable: 0x5555555a3000warning: (Internal error: pc 0x5555555a3000 in read in CU, but not in symtab.) warning: (Error: pc 0x5555555a3000 in address map, but not in symtab.) } $5 = &(i32, i32, [i32]) [(0, 1, 0), (2, 3, 0)] $6 = &(i32, i32, dyn core::fmt::Debug) {pointer: 0x5555555a3020, vtable: 0x5555555a3030warning: (Internal error: pc 0x5555555a3030 in read in CU, but not in symtab.) warning: (Error: pc 0x5555555a3030 in address map, but not in symtab.) } ---->8---- ----8<---- good (gdb 12.1-4 from bookworm) Breakpoint 1 at 0x81dd: file src/main.rs, line 88. [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Breakpoint 1, rustc_gdb_1031745::main () at src/main.rs:88 88 zzz(); // #break $1 = &rustc_gdb_1031745::Foo<[u8]> {data_ptr: 0x7fffffffdd88, length: 4} $2 = &rustc_gdb_1031745::Foo<rustc_gdb_1031745::Foo<[u8]>> {data_ptr: 0x7fffffffdd88, length: 4} $3 = &rustc_gdb_1031745::Foo<dyn core::fmt::Debug> {pointer: 0x555555593034, vtable: 0x5555555a3000} $4 = alloc::boxed::Box<rustc_gdb_1031745::Foo<dyn core::fmt::Debug>, alloc::alloc::Global> {pointer: 0x5555555a7ba0, vtable: 0x5555555a3000} $5 = &(i32, i32, [i32]) {data_ptr: 0x555555593038, length: 2} $6 = &(i32, i32, dyn core::fmt::Debug) {pointer: 0x5555555a3020, vtable: 0x5555555a3030} ---->8---- Since this no longer (fatally) affects the build of rustc, I opted for a lower severity this time around. It looks like an issue with symbol mapping/lookup again. Feel free to adjust if you consider this blocking. 0: https://salsa.debian.org/fg/rustc-gdb-1031745 I am attaching last git reproducer to the bug
Created attachment 14721 [details] test case with readme (v2) - one last issue after proposed patch
(In reply to HectorOron from comment #11) > Since this no longer (fatally) affects the build of rustc, I opted for a > lower severity this time around. It looks like an issue with symbol > mapping/lookup again. Feel free to adjust if you consider this blocking. Would you mind submitting a separate bug for this?
Marking for 13.2
*** Bug 30224 has been marked as a duplicate of this bug. ***
The master branch has been updated by Tom Tromey <tromey@sourceware.org>: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=5f056fcb3dce947447063f5ab225042177a59722 commit 5f056fcb3dce947447063f5ab225042177a59722 Author: Tom Tromey <tromey@adacore.com> Date: Fri Feb 24 10:40:16 2023 -0700 Fix crash in inside_main_func gdb 13.1 crashes while running the rust compiler's debugger tests. The crash has a number of causes. First, the rust compiler still uses the C++-like _Z mangling, but with its own twist -- some hex digits added to the end of a symbol. So, while gdb finds the correct name of "main": (top-gdb) p name $13 = 0x292e0c0 "rustc_gdb_1031745::main" It isn't found in the minsyms, because C++ demangling yields: [99] t 0x90c0 _ZN17rustc_gdb_10317454main17h5b5be7fe16a97225E section .text rustc_gdb_1031745::main::h5b5be7fe16a97225 zko06yobckx336v This could perhaps be fixed. I also filed a new PR to suggest preferring the linkage name of the main program. Next, the rust compiler emits both a DW_TAG_subprogram and a DW_TAG_namespace for "main". This happens because the file is named "main.rs" -- i.e., the bug is specific to the source file name. The crash also seems to require the nested function inside of 'main', at least for me. The namespace always is generated, but perhaps this changes the ordering in the DWARF. When inside_main_func looks up the main symbol, it finds the namespace symbol rather than the function. (I filed a bug about fixing gdb's symbol tables -- long overdue.) Meanwhile, as I think it's important to fix this crash sooner rather than later, this patch changes inside_main_func to check that the symbol that is found is LOC_BLOCK. This perhaps should have been done in the first place, anyway. Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30158
The gdb-13-branch branch has been updated by Tom Tromey <tromey@sourceware.org>: https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=562403d4f5483e0787219fdb88da72c707c8d7f6 commit 562403d4f5483e0787219fdb88da72c707c8d7f6 Author: Tom Tromey <tromey@adacore.com> Date: Fri Feb 24 10:40:16 2023 -0700 Fix crash in inside_main_func gdb 13.1 crashes while running the rust compiler's debugger tests. The crash has a number of causes. First, the rust compiler still uses the C++-like _Z mangling, but with its own twist -- some hex digits added to the end of a symbol. So, while gdb finds the correct name of "main": (top-gdb) p name $13 = 0x292e0c0 "rustc_gdb_1031745::main" It isn't found in the minsyms, because C++ demangling yields: [99] t 0x90c0 _ZN17rustc_gdb_10317454main17h5b5be7fe16a97225E section .text rustc_gdb_1031745::main::h5b5be7fe16a97225 zko06yobckx336v This could perhaps be fixed. I also filed a new PR to suggest preferring the linkage name of the main program. Next, the rust compiler emits both a DW_TAG_subprogram and a DW_TAG_namespace for "main". This happens because the file is named "main.rs" -- i.e., the bug is specific to the source file name. The crash also seems to require the nested function inside of 'main', at least for me. The namespace always is generated, but perhaps this changes the ordering in the DWARF. When inside_main_func looks up the main symbol, it finds the namespace symbol rather than the function. (I filed a bug about fixing gdb's symbol tables -- long overdue.) Meanwhile, as I think it's important to fix this crash sooner rather than later, this patch changes inside_main_func to check that the symbol that is found is LOC_BLOCK. This perhaps should have been done in the first place, anyway. Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30158 (cherry picked from commit 5f056fcb3dce947447063f5ab225042177a59722)
Fixed.
*** Bug 29298 has been marked as a duplicate of this bug. ***