diff options
| author | Jan Beulich <jbeulich@suse.com> | 2021-06-08 14:55:56 +0200 |
|---|---|---|
| committer | Jan Beulich <jbeulich@suse.com> | 2021-06-08 14:55:56 +0200 |
| commit | 77c59789070949a1cfbd19578c6935ccd57869c2 (patch) | |
| tree | 283571b167bf04fce848bd7e472f0ceea993f3ed /gas/testsuite/gas/i386/i386.exp | |
| parent | f185acddfa246009e72e5bea95cd457ccc0cc763 (diff) | |
| download | binutils-gdb-77c59789070949a1cfbd19578c6935ccd57869c2.tar.gz | |
x86-64: avoid bogus warnings with 32-bit addressing
With optimize_disp() adjusting i.types[].bitfield.disp after adjusting
the value to be used as displacement, it better also stores the updated
value, to avoid subsequent "... shortened to ..." warnings. Note how
optimize_imm() already does so.
The -0xffffffff tests being added expose a separate issue: The encoding
chosen should be 1 for ModR/M.mod, not 2. This will want to be taken
care of, but not right here.
This at the same time addresses a similar warning and demonstrates a
similar encoding issue with 16-bit addressing. Since it was omitted
when introducing the lea16-optimize test, add a plain lea16 one to also
cover this.
Diffstat (limited to 'gas/testsuite/gas/i386/i386.exp')
| -rw-r--r-- | gas/testsuite/gas/i386/i386.exp | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/gas/testsuite/gas/i386/i386.exp b/gas/testsuite/gas/i386/i386.exp index db52df4152f..ff88bae92b2 100644 --- a/gas/testsuite/gas/i386/i386.exp +++ b/gas/testsuite/gas/i386/i386.exp @@ -69,6 +69,7 @@ if [gas_32_check] then { run_dump_test "prefix" run_list_test "prefix32" "-al" run_dump_test "lea" + run_dump_test "lea16" run_dump_test "amd" run_dump_test "katmai" run_dump_test "jump" |