lightning
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Lightning] PATCH - Correct comparison of float branch test on i386


From: Paulo César Pereira de Andrade
Subject: [Lightning] PATCH - Correct comparison of float branch test on i386
Date: Wed, 18 Aug 2010 03:35:05 -0300
User-agent: SquirrelMail/1.4.19

  Hi Paolo,

  Can you please review and apply the relevant patch.

[I am not sure about the policy for commit messages and/or
 changelogs but if you give me some pointer, in next messages
 I try to get it right :-), but I only want it fixed, don't
 need my name associated with the patch or something]

  The macro arguments are always bytes, so, I believe the call to
CMPLir must have been a typo.

  This allows at least beqr_f (an alias to beqr_d) and beqr_d to
work on i386 when compiled with -D_ASM_SAFETY.

  In my current test tool, and initial tests, I have this:

/* currently not in i386 if compiled with -D_ASM_SAFETY=1 */
#define FLOAT_BRANCH_WORKS      0
#if FLOAT_BRANCH_WORKS
#  define float_compare_and_branch(LABEL, R0, F0, F1)   \
        beqr_f  LABEL   %F0 %F1
#  define double_compare_and_branch(LABEL, R0, F0, F1)  \
        beqr_d  LABEL   %F0 %F1
#else
#  define float_compare_and_branch(LABEL, R0, F0, F1)   \
        eqr_f %R0 %F0 %F1                               \
        bnei_i LABEL %R0 0
#  define double_compare_and_branch(LABEL, R0, F0, F1)  \
        eqr_d %R0 %F0 %F1                               \
        bnei_i LABEL %R0 0
#endif

but with the patch attached, it works either way.

Thanks,
Paulo

Attachment: 0001-Correct-comparison-of-float-branch-test-on-i386.patch
Description: Binary data


reply via email to

[Prev in Thread] Current Thread [Next in Thread]