genksyms: fix gperf removal conversion
authorLinus Torvalds <torvalds@linux-foundation.org>
Fri, 8 Sep 2017 21:32:34 +0000 (14:32 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 8 Sep 2017 21:32:34 +0000 (14:32 -0700)
commit3aea311c1b4002bd730a09530f80f2f2ad3bf495
treea979c9d7dcc4d7f75f6417e437277b89424f3fc8
parent015a9e66b9b8c1f28097ed09bf9350708e26249a
genksyms: fix gperf removal conversion

I had stupidly missed one special use of 'is_reserved_word()' when I
converted the code to avoid gperf.

I had changed that function to return the token ID directly rather than
a pointer to the token descriptor structure, but that meant that the
test for "is this a reserved word" changed from checking the return
value against NULL, to checking that it wasn't negative.

And while I had converted the main token parser over, I missed the
special case of the typeof phrase handling.  And since our dependency
chain for genksyms does not include the genksyms program itself
changing, my kernel rebuild didn't show the problem.

Fixes: bb3290d91695 ("Remove gperf usage from toolchain")
Reported-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
scripts/genksyms/lex.l
scripts/genksyms/lex.lex.c_shipped