diff options
author | Jan Beulich <jbeulich@suse.com> | 2024-07-26 08:00:42 +0200 |
---|---|---|
committer | Jan Beulich <jbeulich@suse.com> | 2024-07-26 08:00:42 +0200 |
commit | e0b9535c824deb24e209acddcb7662124aeb413f (patch) | |
tree | 5a02d18ed2ba0d4f8d7fee385829cffb2f59d48a /gas/sb.c | |
parent | c97f0d71ea9cc5c3bf63aacef34b5ebc95dd3b9a (diff) | |
download | binutils-e0b9535c824deb24e209acddcb7662124aeb413f.zip binutils-e0b9535c824deb24e209acddcb7662124aeb413f.tar.gz binutils-e0b9535c824deb24e209acddcb7662124aeb413f.tar.bz2 |
gas: drop scrubber state -2
Instead re-use code handling LEX_IS_TWOCHAR_COMMENT_1ST, thus ensuring
that we wouldn't get bogus state transitions: For example, when we're in
states 0 or 1, a comment should be no different from whitespace
encountered in those states. Plus for e.g. x86 this results in such
comments now truly being converted to a blank, as mandated by
documentation. Both aspects apparently were a result of blindly (and
wrongly) moving to state 3 _before_ consuming the "ungot" blank.
Also amend a related comment elsewhere.
In the new testcase the .irp is to make visible in the listing all the
whitespace that the scrubber inserts / leaves in place.
Diffstat (limited to 'gas/sb.c')
0 files changed, 0 insertions, 0 deletions