From bf347fb8e07b175e09c8c3ed3fcb71ecdbdbeff3 Mon Sep 17 00:00:00 2001 From: Nicholas Nethercote Date: Tue, 30 Apr 2002 12:46:22 +0000 Subject: [PATCH] Added note about bts/btr/btc causing bogus cache read counts. Added to todo list note about files with > 65536 lines. git-svn-id: svn://svn.valgrind.org/valgrind/trunk@175 --- cachegrind/docs/manual.html | 9 +++++++++ coregrind/docs/manual.html | 9 +++++++++ docs/manual.html | 9 +++++++++ memcheck/docs/manual.html | 9 +++++++++ 4 files changed, 36 insertions(+) diff --git a/cachegrind/docs/manual.html b/cachegrind/docs/manual.html index a97c2f9fec..4da22db114 100644 --- a/cachegrind/docs/manual.html +++ b/cachegrind/docs/manual.html @@ -2374,6 +2374,14 @@ Valgrind's cache profiling has a number of shortcomings:
  • It doesn't account for cache misses not visible at the instruction level, eg. those arising from TLB misses, or speculative execution.

    • + +

  • The instructions bts, btr and btc + will incorrectly be counted as doing a data read if both the arguments + are registers, eg: + +
    btsl %eax, %edx
    + + This should only happen rarely. Another thing worth nothing is that results are very sensitive. Changing the @@ -2394,6 +2402,7 @@ hopefully they should be close enough to be useful.

  • Program start-up/shut-down calls a lot of functions that aren't interesting and just complicate the output. Would be nice to exclude these somehow.

    • +

  • Handle files with >65535 lines

    • diff --git a/coregrind/docs/manual.html b/coregrind/docs/manual.html index a97c2f9fec..4da22db114 100644 --- a/coregrind/docs/manual.html +++ b/coregrind/docs/manual.html @@ -2374,6 +2374,14 @@ Valgrind's cache profiling has a number of shortcomings:
  • It doesn't account for cache misses not visible at the instruction level, eg. those arising from TLB misses, or speculative execution.

    • + +

  • The instructions bts, btr and btc + will incorrectly be counted as doing a data read if both the arguments + are registers, eg: + +
    btsl %eax, %edx
    + + This should only happen rarely. Another thing worth nothing is that results are very sensitive. Changing the @@ -2394,6 +2402,7 @@ hopefully they should be close enough to be useful.

  • Program start-up/shut-down calls a lot of functions that aren't interesting and just complicate the output. Would be nice to exclude these somehow.

    • +

  • Handle files with >65535 lines

    • diff --git a/docs/manual.html b/docs/manual.html index a97c2f9fec..4da22db114 100644 --- a/docs/manual.html +++ b/docs/manual.html @@ -2374,6 +2374,14 @@ Valgrind's cache profiling has a number of shortcomings:
  • It doesn't account for cache misses not visible at the instruction level, eg. those arising from TLB misses, or speculative execution.

    • + +

  • The instructions bts, btr and btc + will incorrectly be counted as doing a data read if both the arguments + are registers, eg: + +
    btsl %eax, %edx
    + + This should only happen rarely. Another thing worth nothing is that results are very sensitive. Changing the @@ -2394,6 +2402,7 @@ hopefully they should be close enough to be useful.

  • Program start-up/shut-down calls a lot of functions that aren't interesting and just complicate the output. Would be nice to exclude these somehow.

    • +

  • Handle files with >65535 lines

    • diff --git a/memcheck/docs/manual.html b/memcheck/docs/manual.html index a97c2f9fec..4da22db114 100644 --- a/memcheck/docs/manual.html +++ b/memcheck/docs/manual.html @@ -2374,6 +2374,14 @@ Valgrind's cache profiling has a number of shortcomings:
  • It doesn't account for cache misses not visible at the instruction level, eg. those arising from TLB misses, or speculative execution.

    • + +

  • The instructions bts, btr and btc + will incorrectly be counted as doing a data read if both the arguments + are registers, eg: + +
    btsl %eax, %edx
    + + This should only happen rarely. Another thing worth nothing is that results are very sensitive. Changing the @@ -2394,6 +2402,7 @@ hopefully they should be close enough to be useful.

  • Program start-up/shut-down calls a lot of functions that aren't interesting and just complicate the output. Would be nice to exclude these somehow.

    • +

  • Handle files with >65535 lines

    • -- 2.47.2