|
From: Paul M. <le...@us...> - 2006-08-03 21:43:09
|
Update of /cvsroot/linuxsh/linux/arch/sh/lib In directory sc8-pr-cvs6.sourceforge.net:/tmp/cvs-serv12384/arch/sh/lib Modified Files: memcpy-sh4.S Log Message: Cleanup whitespace damage. Index: memcpy-sh4.S =================================================================== RCS file: /cvsroot/linuxsh/linux/arch/sh/lib/memcpy-sh4.S,v retrieving revision 1.2 retrieving revision 1.3 diff -u -d -r1.2 -r1.3 --- memcpy-sh4.S 3 Aug 2006 12:35:47 -0000 1.2 +++ memcpy-sh4.S 3 Aug 2006 21:42:23 -0000 1.3 @@ -88,7 +88,6 @@ 9: rts nop - ! ! GHIJ KLMN OPQR --> .GHI JKLM NOPQ R... @@ -192,15 +191,16 @@ ! Arguments are not nicely long word aligned or zero len. ! Check for small copies, and if so do a simple byte at a time copy. ! - ! Deciding on an exact value of 'small' is not easy, as the point at which - ! using the optimised routines become worthwhile varies (these are the - ! cycle counts for differnet sizes using byte-at-a-time vs. optimised): + ! Deciding on an exact value of 'small' is not easy, as the point at + ! which using the optimised routines become worthwhile varies (these + ! are the cycle counts for differnet sizes using byte-at-a-time vs. + ! optimised): ! size byte-at-time long word byte ! 16 42 39-40 46-50 50-55 ! 24 58 43-44 54-58 62-67 ! 36 82 49-50 66-70 80-85 - ! However the penalty for getting it 'wrong' is much higher for long word - ! aligned data (and this is more common), so use a value of 16. + ! However the penalty for getting it 'wrong' is much higher for long + ! word aligned data (and this is more common), so use a value of 16. cmp/gt r6,r1 ! 56 MT @@ -393,7 +393,8 @@ 9: rts nop - ! Size is at least 64 bytes, so will be going round the big loop at least once. + ! Size is at least 64 bytes, so will be going round the + ! big loop at least once. ! ! r2 = rounded up r4 ! r3 = rounded down r0 @@ -595,7 +596,8 @@ .balign 32 .Lcase2b: - ! Size is at least 64 bytes, so will be going round the big loop at least once. + ! Size is at least 64 bytes, so will be going round the + ! big loop at least once. ! ! r2 = rounded up r4 ! r3 = rounded down r0 @@ -729,7 +731,7 @@ xtrct r9, r8 ! 48 EX mov.l @(0x00,r5), r12 ! 18 LS (latency=2) - xtrct r10, r9 ! 48 EX + xtrct r10, r9 ! 48 EX movca.l r0,@r1 ! 40 LS (latency=3-7) add #-0x1c, r1 ! 50 EX @@ -797,4 +799,3 @@ mov.b @(r0,r5),r1 rts mov.b r1,@-r0 - |