Enable heap profiling on macOS #2314
Conversation
m-seker
force-pushed
the
osx_heap_profiling
branch
4 times, most recently
from
d508d10
to
42d1314
Compare
| @@ -4698,7 +4698,7 @@ sub ParseLibraries { | |||
| $offset = HexExtend($3); | |||
| $lib = $4; | |||
| $lib =~ s|\\|/|g; # turn windows-style paths into unix-style paths | |||
| } elsif ($l =~ /^\s*($h)-($h):\s*(\S+\.so(\.\d+)*)/) { | |||
| } elsif ($l =~ /^\s*($h)-($h):\s*(.*$)/) { | |||
There was a problem hiding this comment.
Match for library mappings on macOS dump. The library names can have .dylib extension or no extension at all.
There was a problem hiding this comment.
A bit concerned that the regex is too relaxed. Any idea to tighten the no extension case a bit? It looks space is allowed on macOS.
| @@ -1369,7 +1380,7 @@ fi | |||
| if test "x$backtrace_method" = "x" -a "x$enable_prof_libgcc" = "x1" \ | |||
| -a "x$GCC" = "xyes" ; then | |||
| AC_CHECK_HEADERS([unwind.h], , [enable_prof_libgcc="0"]) | |||
| if test "x${enable_prof_libgcc}" = "x1" ; then | |||
| if test "x${enable_prof_libgcc}" = "x1" -a "x${je_cv_clang}" != "xyes" ; then | |||
There was a problem hiding this comment.
No need to link with lgcc for clang.
There was a problem hiding this comment.
Shall we still perform the _Unwind_Backtrace availability test for clang? Or it's for sure present with unwind.h?
m-seker
force-pushed
the
osx_heap_profiling
branch
from
42d1314
to
46455aa
Compare
| @@ -4698,7 +4698,7 @@ sub ParseLibraries { | |||
| $offset = HexExtend($3); | |||
| $lib = $4; | |||
| $lib =~ s|\\|/|g; # turn windows-style paths into unix-style paths | |||
| } elsif ($l =~ /^\s*($h)-($h):\s*(\S+\.so(\.\d+)*)/) { | |||
| } elsif ($l =~ /^\s*($h)-($h):\s*(.*$)/) { | |||
There was a problem hiding this comment.
A bit concerned that the regex is too relaxed. Any idea to tighten the no extension case a bit? It looks space is allowed on macOS.
| @@ -1369,7 +1380,7 @@ fi | |||
| if test "x$backtrace_method" = "x" -a "x$enable_prof_libgcc" = "x1" \ | |||
| -a "x$GCC" = "xyes" ; then | |||
| AC_CHECK_HEADERS([unwind.h], , [enable_prof_libgcc="0"]) | |||
| if test "x${enable_prof_libgcc}" = "x1" ; then | |||
| if test "x${enable_prof_libgcc}" = "x1" -a "x${je_cv_clang}" != "xyes" ; then | |||
There was a problem hiding this comment.
Shall we still perform the _Unwind_Backtrace availability test for clang? Or it's for sure present with unwind.h?
| typedef int (prof_dump_open_maps_t)(); | ||
| extern prof_dump_open_maps_t *JET_MUTABLE prof_dump_open_maps; | ||
| #endif // __APPLE__ |
| typedef struct segment_command_64 je_dyld_segment_command; | ||
|
|
||
| #define JE_DYLD_IMAGE_MAGIC MH_MAGIC_64 | ||
| #define JE_DYLD_IMAGE_MAGIC_INV MH_CIGAM_64 |
There was a problem hiding this comment.
Maybe _REVERSE or stick to _CIGAM? The _INV feels a bit close to invalid.
|
|
||
| TEST_BEGIN(test_mdump_maps_error) { | ||
| test_skip_if(!config_prof); | ||
| test_skip_if(!config_debug); | ||
|
|
||
| prof_dump_open_file_t *open_file_orig = prof_dump_open_file; | ||
| prof_dump_write_file_t *write_file_orig = prof_dump_write_file; | ||
| #ifndef __APPLE__ |
There was a problem hiding this comment.
nit: It looks we can remove this too?
| @@ -128,10 +128,12 @@ TEST_BEGIN(test_mdump_output_error) { | |||
| } | |||
| TEST_END | |||
|
|
|||
| #ifndef __APPLE__ | |||
Enable heap profiling on macOS.
Both prof-libgcc and prof-gcc works with clang on macOS. The library mappings are dumped via dyld API.