This is the mail archive of the
systemtap@sourceware.org
mailing list for the systemtap project.
[Bug translator/13100] Building tracepoint cache all at once fails when kernel debuginfo installed
- From: "jistone at redhat dot com" <sourceware-bugzilla at sourceware dot org>
- To: systemtap at sourceware dot org
- Date: Wed, 05 Oct 2011 19:24:45 +0000
- Subject: [Bug translator/13100] Building tracepoint cache all at once fails when kernel debuginfo installed
- Auto-submitted: auto-generated
- References: <bug-13100-6586@http.sourceware.org/bugzilla/>
http://sourceware.org/bugzilla/show_bug.cgi?id=13100
Josh Stone <jistone at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jistone at redhat dot com
--- Comment #1 from Josh Stone <jistone at redhat dot com> 2011-10-05 19:24:45 UTC ---
The batch tracequery, with all discovered headers at once, was something of an
optimization, and experience has shown it to be too optimistic. This idea was
so we could have only one kbuild make invocation (which is fairly expensive)
and only one resulting tracequery.ko to juggle. But the tracepoint headers so
often conflict with each other that we almost always end up falling back to
separate tracequery.ko per header.
So if we accept that as a fact of life, then perhaps we should just restructure
it to always do separate tracequery.ko, but we could combine them into a single
Makefile for then only one kbuild invocation. That should hopefully relieve
much of the overhead of building them separately, even though they'll still be
forking lots of compiler instances.
If we can find a way to soft-fail any unusable headers, then these objects
could still be combined into one big tracequery.ko. That would make cache
management easier, but I'm not sure if kbuild can express this easily.
--
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.