varun-r-mallya/sysprof
1
0
Fork 0
mirror of https://github.com/varun-r-mallya/sysprof.git synced 2026-02-12 08:00:53 +00:00
Code Issues Packages Projects Releases Wiki Activity

TODO

Browse Source
This commit is contained in:
Søren Sandmann Pedersen
2005-10-31 04:20:25 +00:00
parent 60757b6977
commit c34a9b862f
2 changed files with 81 additions and 52 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
TODO
View File

@ -23,7 +23,7 @@ Before 1.2:
esp0 points to top of kernel stack esp0 points to top of kernel stack
esp points to top of user stack esp points to top of user stack
(Reported by Kjartan Marass). (Reported by Kjartan Maraas).
- Fix bugs/performance issues: - Fix bugs/performance issues:
- total should probably be cached so that compute_total() doesn't - total should probably be cached so that compute_total() doesn't
@ -58,11 +58,38 @@ Before 1.2:
- don't leak the presentation strings/objects - don't leak the presentation strings/objects
- loading and saving probably leak right now - loading and saving probably leak right now
- think about loading and saving. Goals - rethink loading and saving. Goals
- Can load 1.0 profiles - Can load 1.0 profiles
- Don't export too much of stackstashes to the rest of the - Don't export too much of stackstashes to the rest of the
app app
Features:
* Have a compatibility module that can load 1.0 modules
- reads in the call tree, then generate a stack stash
by repeatedly adding traces.
* Make stackstash able to save themselves using callbacks.
* If loading a file fails, then try loading it using the
1.0 loader.
- optimization: make sure we fail immediately if we
see an unknown object.
* Add versioning support to sfile.[ch]:
- sformat_new() should take doctype and version strings:
like "sysprof-profile" "version 1.2"
- there should be sfile_sniff() functionality that will
return the doctype and version of a file. Or None
if there aren't any.
* At this point, make the loader first check if the file has a version
if it doesn't, load as 1.0, otherwise as whatever the version claims
it is.
* Make provisions for forward compatibility: maybe it should be possible
to load records with more fields than specified.
* Figure out how to make sfile.[ch] use less memory. * Figure out how to make sfile.[ch] use less memory.
- In general clean sfile.[ch] up a little: - In general clean sfile.[ch] up a little:
- split out dfa in its own generic class - split out dfa in its own generic class
@ -88,8 +115,11 @@ Before 1.2:
kernel stack and user space stack, have userspace stitch them kernel stack and user space stack, have userspace stitch them
together. well, they could be stitched together in the kernel. together. well, they could be stitched together in the kernel.
Already done: we now take a stacktrace of the user space process Already done: we now take a stacktrace of the user space process
when the interrupt happens in kernel mode. We don't take any when the interrupt happens in kernel mode. (Unfortunately, this
stacktraces of the kernel though. Things that need to be investigated: causes lockups on many kernels).
We don't take any stacktraces of the kernel though. Things that need to be
investigated:
- does the kernel come with dwarf debug information? - does the kernel come with dwarf debug information?
- does the kernel come with some other debug info - does the kernel come with some other debug info
- is there a place where the vmlinux binary is usually - is there a place where the vmlinux binary is usually

1
profile.c
View File

@ -530,7 +530,6 @@ profile_list_callers (Profile *profile,
g_hash_table_destroy (callers_by_object); g_hash_table_destroy (callers_by_object);
return result; return result;
} }
void void