From a37739c10ee16f9972fd059486971b572a67b545 Mon Sep 17 00:00:00 2001 From: Kazu Hirata Date: Tue, 16 Nov 2004 23:26:21 +0000 Subject: [PATCH] * expmed.c, ra-colorize.c: Fix comment typos. From-SVN: r90771 --- gcc/ChangeLog | 4 ++++ gcc/expmed.c | 2 +- gcc/ra-colorize.c | 2 +- 3 files changed, 6 insertions(+), 2 deletions(-) diff --git a/gcc/ChangeLog b/gcc/ChangeLog index 600089338f6..da8619d6bb0 100644 --- a/gcc/ChangeLog +++ b/gcc/ChangeLog @@ -1,3 +1,7 @@ +2004-11-16 Kazu Hirata + + * expmed.c, ra-colorize.c: Fix comment typos. + 2004-11-16 Daniel Jacobowitz Mark Mitchell diff --git a/gcc/expmed.c b/gcc/expmed.c index b2e62350b1d..033d94952be 100644 --- a/gcc/expmed.c +++ b/gcc/expmed.c @@ -2555,7 +2555,7 @@ synth_mult (struct algorithm *alg_out, unsigned HOST_WIDE_INT t, /* If the target has a cheap shift-and-add instruction use that in preference to a shift insn followed by an add insn. Assume that the shift-and-add is "atomic" with a latency - equal to it's cost, otherwise assume that on superscalar + equal to its cost, otherwise assume that on superscalar hardware the shift may be executed concurrently with the earlier steps in the algorithm. */ op_cost = add_cost[mode] + shift_cost[mode][m]; diff --git a/gcc/ra-colorize.c b/gcc/ra-colorize.c index 6d7980cc2b3..08bd960d24c 100644 --- a/gcc/ra-colorize.c +++ b/gcc/ra-colorize.c @@ -1323,7 +1323,7 @@ colorize_one_web (struct web *web, int hard) usable_regs of this web (which is probably larger than that of the preferred or alternate class). All searches first try to find a non-call-clobbered hard-reg. - XXX this should be more finegraned... First look into preferred + XXX this should be more fine grained... First look into preferred non-callclobbered hardregs, then _if_ the web crosses calls, in alternate non-cc hardregs, and only _then_ also in preferred cc hardregs (and alternate ones). Currently we don't track the number -- 2.30.2