summaryrefslogtreecommitdiffstats
path: root/dwflpp.cxx
diff options
context:
space:
mode:
authorJosh Stone <jistone@redhat.com>2009-06-08 15:36:42 -0700
committerJosh Stone <jistone@redhat.com>2009-06-08 15:48:33 -0700
commit0e14e0793ffb891bccd465cf518480685e3cd49e (patch)
treebf35be7cb408dc8aaa78037b0ca78c48bcb4dd89 /dwflpp.cxx
parenta77245e55075e1b756b51a5226483fe5e618940e (diff)
downloadsystemtap-steved-0e14e0793ffb891bccd465cf518480685e3cd49e.tar.gz
systemtap-steved-0e14e0793ffb891bccd465cf518480685e3cd49e.tar.xz
systemtap-steved-0e14e0793ffb891bccd465cf518480685e3cd49e.zip
Let query_module abort early for simple matches
query_module was already returning DW_CB_ABORT when a simple match was found, but dwflpp::iterate_over_modules was ignoring that and instead forcing the module loop to restart. The only way out of the loop was with the pending_interrupts flag, which is only for signalled interrupts. Now iterate_over_modules will only attempt the dwfl_getmodules loop once, since that loop will only abort if the CB returns DW_CB_ABORT. Then query_module is also modified to return ABORT if pending_interrupts is flagged. My trusty test, stap -l syscall.*, is nearly 2x faster with this change. Empirically, I found that the kernel object is always the first "module" returned, so the syscall probepoints always gets to short-circuit the loop right away.
Diffstat (limited to 'dwflpp.cxx')
-rw-r--r--dwflpp.cxx9
1 files changed, 2 insertions, 7 deletions
diff --git a/dwflpp.cxx b/dwflpp.cxx
index 3af26053..1a589771 100644
--- a/dwflpp.cxx
+++ b/dwflpp.cxx
@@ -397,13 +397,8 @@ dwflpp::iterate_over_modules(int (* callback)(Dwfl_Module *, void **,
void *),
base_query *data)
{
- ptrdiff_t off = 0;
- do
- {
- if (pending_interrupts) return;
- off = dwfl_getmodules (dwfl, callback, data, off);
- }
- while (off > 0);
+ dwfl_getmodules (dwfl, callback, data, 0);
+
// Don't complain if we exited dwfl_getmodules early.
// This could be a $target variable error that will be
// reported soon anyway.