[Bug ld/29512] ld non-canon ref to canon protected function check breaks Solaris/x86
https://sourceware.org/bugzilla/show_bug.cgi?id=29512 Florian Weimer changed: What|Removed |Added CC||fweimer at redhat dot com -- You are receiving this mail because: You are on the CC list for the bug.
Segmentation Fault on (%dx) [x86]
GNU assembler (GNU Binutils) 2.39.50.20221104 Configured for target x86_64-pc-linux-gnu Compiled through GCC (12.1.0) The bug causes a segmentation fault whenever (at least for the instructions I tried it with) assembler attempts to assemble the instruction with indirect reference to memory through %dx, that is, one of the operands is (%dx). The dereference is invalid and thus should be reported as such, but it instead segfaults The reason for segmentation fault lies in a null pointer dereference, that is caused by possibly an incorrect parse of the instruction by `parse_insn` For example, attempting to assemble ``` _start : movw (%dx), %dx ``` causes a segfault at `config/tc-i386.c`, where attempt is made to dereference `i.op[0]`'s `regs` field, which happens to be NULL Whenever an instruction is incorrectly encoded, it generates an error when parsing the instructions (`parse_insn`) which does not happen here, and wrongly parses information about operands (such as `i.mem_operands` being 0) Since I spent a significant amount of time looking into it, I would like to fix this bug. I would appreciate any suggestions, pointers to documentations, etc that may be helpful to me It may have to do with some special case associated with (%dx) but I will have to look further into it before I can reach any conclusion and make a fix
[Bug gas/29751] New: (%dx) is mishandled
https://sourceware.org/bugzilla/show_bug.cgi?id=29751 Bug ID: 29751 Summary: (%dx) is mishandled Product: binutils Version: 2.40 (HEAD) Status: NEW Severity: normal Priority: P2 Component: gas Assignee: unassigned at sourceware dot org Reporter: hjl.tools at gmail dot com Target Milestone: --- Target: i386 The memory operand of (%dx) should be only used with input/output instructions. But we failed to properly check it: [hjl@gnu-tgl-3 tmp]$ cat a1.s incl (%dx) [hjl@gnu-tgl-3 tmp]$ gcc -c a1.s [hjl@gnu-tgl-3 tmp]$ objdump -dw a1.o a1.o: file format elf64-x86-64 Disassembly of section .text: <.text>: 0: ff 00 incl (%rax) [hjl@gnu-tgl-3 tmp]$ cat a2.s mov (%dx), %ax [hjl@gnu-tgl-3 tmp]$ gcc -c a2.s a2.s: Assembler messages: a2.s:1: Internal error (Segmentation fault). Please report this bug. [hjl@gnu-tgl-3 tmp]$ -- You are receiving this mail because: You are on the CC list for the bug.
[Bug binutils/26160] [2.35 Regression] Warning: index table size is too small 0 vs 70
https://sourceware.org/bugzilla/show_bug.cgi?id=26160 --- Comment #5 from cvs-commit at gcc dot gnu.org --- The master branch has been updated by H.J. Lu : https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=d195b95c91febcdad3ec002d415586cee7f6cb16 commit d195b95c91febcdad3ec002d415586cee7f6cb16 Author: H.J. Lu Date: Mon Oct 31 17:02:17 2022 -0700 binutils: Run PR binutils/26160 test Update expected PR binutils/26160 test output for readelf out change and run PR binutils/26160 test. PR binutils/26160 * testsuite/binutils-all/pr26160.r: Updated. * testsuite/binutils-all/readelf.exp: Run PR binutils/26160 test. -- You are receiving this mail because: You are on the CC list for the bug.
