diff options
author | Alan Modra <amodra@gmail.com> | 2020-01-13 14:27:19 +1030 |
---|---|---|
committer | Alan Modra <amodra@gmail.com> | 2020-01-13 16:44:27 +1030 |
commit | febda64f152972d8edf0147fb29f89b02f6a4cf9 (patch) | |
tree | 530633bd36c29a3fd0715e8658777f665385aa9a /opcodes/m32r-desc.h | |
parent | df08b5881b4972d78f9a2069955dad5b12bc972e (diff) | |
download | gdb-febda64f152972d8edf0147fb29f89b02f6a4cf9.zip gdb-febda64f152972d8edf0147fb29f89b02f6a4cf9.tar.gz gdb-febda64f152972d8edf0147fb29f89b02f6a4cf9.tar.bz2 |
ubsan: wasm32: signed integer overflow
The signed integer overflow occurred when adding one to target_count
for (i = 0; i < target_count + 1; i++)
but that's the least of the worries here. target_count was long and i
int, leading to the possibility of a loop that never ended.
So to avoid this type of vulnerability, this patch uses what I believe
to be the proper types for arguments of various wasm32 opcodes, rather
than using "long" which may change in size.
gas/
* testsuite/gas/wasm32/allinsn.d: Update expected output.
opcodes/
* wasm32-dis.c (print_insn_wasm32): Localise variables. Store
result of wasm_read_leb128 in a uint64_t and check that bits
are not lost when copying to other locals. Use uint32_t for
most locals. Use PRId64 when printing int64_t.
Diffstat (limited to 'opcodes/m32r-desc.h')
0 files changed, 0 insertions, 0 deletions