blktrace.git
18 years ago[PATCH] blktrace: better check for mounted relayfs
Nathan Scott [Thu, 15 Sep 2005 07:31:12 +0000 (09:31 +0200)]
[PATCH] blktrace: better check for mounted relayfs

Don't just check for the presence of the directory, also check whether
relayfs really is mounted there.

18 years ago[PATCH] blkparse: add stopwatch functionality
Nathan Scott [Thu, 15 Sep 2005 07:29:24 +0000 (09:29 +0200)]
[PATCH] blkparse: add stopwatch functionality

Like the blktrace -w option, but allow for defining an interval as well.

18 years ago[PATCH] blktrace: add stopwatch functionality
Nathan Scott [Wed, 14 Sep 2005 17:32:27 +0000 (19:32 +0200)]
[PATCH] blktrace: add stopwatch functionality

Enables you to stop the trace after a defined period has passed.

18 years ago[PATCH] btrace: make the output easily parsably by default
Nathan Scott [Wed, 14 Sep 2005 17:30:49 +0000 (19:30 +0200)]
[PATCH] btrace: make the output easily parsably by default

Drop CPU/device summary unless specifically requested.

18 years ago[PATCH] Add btrace to Makefile
Nathan Scott [Wed, 14 Sep 2005 17:30:14 +0000 (19:30 +0200)]
[PATCH] Add btrace to Makefile

18 years ago[PATCH] Add btrace script
Nathan Scott [Wed, 14 Sep 2005 17:29:40 +0000 (19:29 +0200)]
[PATCH] Add btrace script

Using blktrace and blkparse in the "standard" piped setup is a little
cumbersome and not very intuitive, so add a script to do this automatically.
This is what users would expect.

18 years ago[PATCH] Add support for tracing multiple devices
Nathan Scott [Mon, 12 Sep 2005 08:06:59 +0000 (10:06 +0200)]
[PATCH] Add support for tracing multiple devices

This bumps the protocol number to 0x05, as device info was added to the
trace structure.

18 years ago[PATCH] blkparse: Fix mistaken double rb lookup in find_track()
Nathan Scott [Mon, 12 Sep 2005 07:21:00 +0000 (09:21 +0200)]
[PATCH] blkparse: Fix mistaken double rb lookup in find_track()

18 years ago[PATCH] README: document the -t option
Jens Axboe [Fri, 9 Sep 2005 19:10:58 +0000 (21:10 +0200)]
[PATCH] README: document the -t option

18 years ago[PATCH] blkparse: move magic check a little earlier in stdin read
Jens Axboe [Fri, 9 Sep 2005 18:13:55 +0000 (20:13 +0200)]
[PATCH] blkparse: move magic check a little earlier in stdin read

Otherwise we don't know if ->pdu_len and other fields are truly valid.

18 years ago[PATCH] blkparse: fix quit on stdin reads
Jens Axboe [Fri, 9 Sep 2005 15:07:17 +0000 (17:07 +0200)]
[PATCH] blkparse: fix quit on stdin reads

If the blocking read returns 0, we really are done. So safely exit.

18 years ago[PATCH] blkparse: fixup stats
Jens Axboe [Fri, 9 Sep 2005 14:28:14 +0000 (16:28 +0200)]
[PATCH] blkparse: fixup stats

Pretty up the output a little, split into read/write.

18 years ago[PATCH] blkparse: add per-process stats for longest waits
Jens Axboe [Fri, 9 Sep 2005 14:07:39 +0000 (16:07 +0200)]
[PATCH] blkparse: add per-process stats for longest waits

18 years ago[PATCH] blkparse: Add request allocation -> queue time
Jens Axboe [Fri, 9 Sep 2005 13:56:02 +0000 (15:56 +0200)]
[PATCH] blkparse: Add request allocation -> queue time

18 years ago[PATCH] blkparse: change formatting of output a little
Jens Axboe [Fri, 9 Sep 2005 13:47:26 +0000 (15:47 +0200)]
[PATCH] blkparse: change formatting of output a little

18 years ago[PATCH] blkparse: optimize the read_sort_events() memory allocations
Jens Axboe [Fri, 9 Sep 2005 11:36:42 +0000 (13:36 +0200)]
[PATCH] blkparse: optimize the read_sort_events() memory allocations

It's a little silly to malloc/free the buffer all the time, more practical
to let it settle on its own.

18 years agoMerge refs/heads/master from origin
Jens Axboe [Fri, 9 Sep 2005 13:25:47 +0000 (15:25 +0200)]
Merge refs/heads/master from origin

18 years ago[PATCH] blkparse: Display io queue/completion stats
Jens Axboe [Fri, 9 Sep 2005 13:24:13 +0000 (15:24 +0200)]
[PATCH] blkparse: Display io queue/completion stats

Display the time a request spent in the io scheduler and the time it took
that given request to be completed by the drive/driver.

18 years ago[PATCH] blkparse: fix leak in stdin read and sort
Jens Axboe [Fri, 9 Sep 2005 10:30:33 +0000 (12:30 +0200)]
[PATCH] blkparse: fix leak in stdin read and sort

18 years ago[PATCH] blkparse: add framework for tracking individual ios
Jens Axboe [Fri, 9 Sep 2005 10:04:17 +0000 (12:04 +0200)]
[PATCH] blkparse: add framework for tracking individual ios

18 years ago[PATCH] blkparse: change output buffering
Jens Axboe [Fri, 9 Sep 2005 07:25:31 +0000 (09:25 +0200)]
[PATCH] blkparse: change output buffering

If logging to stdout, only use a per-line buffer instead of fully buffered.

18 years ago[PATCH] blktrace: Fix whitespace in log_merge()
Nathan Scott [Fri, 9 Sep 2005 07:22:52 +0000 (09:22 +0200)]
[PATCH] blktrace: Fix whitespace in log_merge()

18 years ago[PATCH] blktrace: remove unneeded strdup
Nathan Scott [Fri, 9 Sep 2005 07:22:13 +0000 (09:22 +0200)]
[PATCH] blktrace: remove unneeded strdup

18 years ago[PATCH] Fix dispatch reporting
Nathan Scott [Fri, 9 Sep 2005 07:20:26 +0000 (09:20 +0200)]
[PATCH] Fix dispatch reporting

18 years ago[PATCH] Fix time typo, it's in nanoseconds
Nathan Scott [Fri, 9 Sep 2005 07:19:26 +0000 (09:19 +0200)]
[PATCH] Fix time typo, it's in nanoseconds

18 years ago[PATCH] Use __WORDSIZE in hash
Jens Axboe [Fri, 9 Sep 2005 07:18:49 +0000 (09:18 +0200)]
[PATCH] Use __WORDSIZE in hash

Originally fron Nathan, improved to remove 'bpl' as well.

18 years ago[PATCH] blkparse: Add per-process stats and fix output
Jens Axboe [Thu, 8 Sep 2005 18:15:18 +0000 (20:15 +0200)]
[PATCH] blkparse: Add per-process stats and fix output

Add the option (-s) to display per-process read/write stats as well as
the per-processor and total output. At the same time, fix the output so
it's a single file instead of per-processor (doesn't make sense here) and
put everything in the output file given with -o. If no output file is given,
stdout is used.

18 years ago[PATCH] Update README with info about the snapshots
Jens Axboe [Thu, 8 Sep 2005 09:34:26 +0000 (11:34 +0200)]
[PATCH] Update README with info about the snapshots

18 years agoMerge refs/heads/master from origin
Jens Axboe [Thu, 8 Sep 2005 06:37:57 +0000 (08:37 +0200)]
Merge refs/heads/master from origin

18 years ago[PATCH] Add patch for Linus' kernel
Jens Axboe [Thu, 8 Sep 2005 06:37:33 +0000 (08:37 +0200)]
[PATCH] Add patch for Linus' kernel

relayfs has been merged now, so it's easier to provide a mainline patch as well.

18 years ago[PATCH] Fix unaligned access on pthread_join(..., &int)
Jens Axboe [Wed, 7 Sep 2005 14:17:07 +0000 (16:17 +0200)]
[PATCH] Fix unaligned access on pthread_join(..., &int)

18 years ago[PATCH] Misc stdin fixes
Jens Axboe [Wed, 7 Sep 2005 11:05:20 +0000 (13:05 +0200)]
[PATCH] Misc stdin fixes

- Allow skipping in the returned stdin data, otherwise I've noticed loss
  of as many as 'nr' events when reading from the network.

- Use realloc() in resize_buffers()

- Detect sort_entries() errors in stdin read

18 years ago[PATCH] Set locale in blkparse
Jens Axboe [Tue, 6 Sep 2005 12:52:04 +0000 (14:52 +0200)]
[PATCH] Set locale in blkparse

Otherwise the 1000 numerator wont work.

18 years ago[PATCH] Correct netcat examples
Jens Axboe [Tue, 6 Sep 2005 12:49:51 +0000 (14:49 +0200)]
[PATCH] Correct netcat examples

18 years ago[PATCH] Minimize output fd lock time
Jens Axboe [Tue, 6 Sep 2005 12:46:45 +0000 (14:46 +0200)]
[PATCH] Minimize output fd lock time

Extract both trace and payload before acquiring the output descriptor lock.
The writes should take very shortly thus minimizing the hold time, the read
may take much longer.

18 years ago[PATCH] Unify error returns
Jens Axboe [Tue, 6 Sep 2005 12:10:33 +0000 (14:10 +0200)]
[PATCH] Unify error returns

It's all a mess of dupes now anyway.

18 years ago[PATCH] Update documentation
Jens Axboe [Tue, 6 Sep 2005 07:34:28 +0000 (09:34 +0200)]
[PATCH] Update documentation

Some of it was a little out of sync with reality. Add an example for live
tracing, locally as well as over the network.

18 years ago[PATCH] Add option to limit rb batching for live traces
Jens Axboe [Tue, 6 Sep 2005 07:09:52 +0000 (09:09 +0200)]
[PATCH] Add option to limit rb batching for live traces

Right now, blkparse keeps going until there's a delay in events read.
Add an option to cap the number of pending entries (-b) and default to
1024 events.

18 years ago[PATCH] Missed the ->device in endianness conversions
Jens Axboe [Tue, 6 Sep 2005 07:01:30 +0000 (09:01 +0200)]
[PATCH] Missed the ->device in endianness conversions

18 years ago[PATCH] Update kernel patch to revision 4 protocol
Jens Axboe [Tue, 6 Sep 2005 07:00:17 +0000 (09:00 +0200)]
[PATCH] Update kernel patch to revision 4 protocol

18 years ago[PATCH] Backout the ->device change from revision 4
Jens Axboe [Tue, 6 Sep 2005 06:57:57 +0000 (08:57 +0200)]
[PATCH] Backout the ->device change from revision 4

18 years ago[PATCH] allow devices as arguments (as well as via -d option)
Nathan Scott [Tue, 6 Sep 2005 06:53:36 +0000 (08:53 +0200)]
[PATCH] allow devices as arguments (as well as via -d option)

18 years ago[PATCH] Pretty up output on large SMP machines
Nathan Scott [Tue, 6 Sep 2005 06:52:44 +0000 (08:52 +0200)]
[PATCH] Pretty up output on large SMP machines

18 years ago[PATCH] Missing colon in some output
Nathan Scott [Tue, 6 Sep 2005 06:51:55 +0000 (08:51 +0200)]
[PATCH] Missing colon in some output

18 years ago[PATCH] Fix per_cpu_info->cpu setting
Jens Axboe [Tue, 6 Sep 2005 06:51:15 +0000 (08:51 +0200)]
[PATCH] Fix per_cpu_info->cpu setting

Not all traces are from CPU 0! Noticed by Nathan.

18 years ago[PATCH] Dump command process names, they are available now
Nathan Scott [Tue, 6 Sep 2005 06:49:15 +0000 (08:49 +0200)]
[PATCH] Dump command process names, they are available now

18 years ago[PATCH] Bump ->cpu to 32-bits for alignment and add process name
Nathan Scott [Tue, 6 Sep 2005 06:48:11 +0000 (08:48 +0200)]
[PATCH] Bump ->cpu to 32-bits for alignment and add process name

18 years ago[PATCH] Fix some wierd pthread/atexit interactions
Nathan Scott [Tue, 6 Sep 2005 06:46:31 +0000 (08:46 +0200)]
[PATCH] Fix some wierd pthread/atexit interactions

18 years ago[PATCH] Forgot to update last file on checkin
Jens Axboe [Tue, 6 Sep 2005 06:25:57 +0000 (08:25 +0200)]
[PATCH] Forgot to update last file on checkin

18 years ago[PATHC] Plug some more stoptrace holes
Nathan Scott [Tue, 6 Sep 2005 06:23:03 +0000 (08:23 +0200)]
[PATHC] Plug some more stoptrace holes

I came across some more places where we can bail out of blktrace while
accidentally leaving the bdev trace running... this patch fixes them.

18 years ago[PATCH] Make per_cpu_info[] dynamically allocated and extendable
Nathan Scott [Tue, 6 Sep 2005 06:21:43 +0000 (08:21 +0200)]
[PATCH] Make per_cpu_info[] dynamically allocated and extendable

18 years ago[PATCH] Missing ->cpu endianness conversion
Nathan Scott [Tue, 6 Sep 2005 06:05:07 +0000 (08:05 +0200)]
[PATCH] Missing ->cpu endianness conversion

18 years ago[PATCH] Misc error handling and style fixes
Jens Axboe [Mon, 5 Sep 2005 17:50:29 +0000 (19:50 +0200)]
[PATCH] Misc error handling and style fixes

18 years ago[PATCH] Move open of logging file
Jens Axboe [Mon, 5 Sep 2005 14:35:10 +0000 (16:35 +0200)]
[PATCH] Move open of logging file

It should not be in the sorting, move it to the dumping place instead.

18 years ago[PATCH] Add support for killing running trace
Jens Axboe [Mon, 5 Sep 2005 14:01:46 +0000 (16:01 +0200)]
[PATCH] Add support for killing running trace

Sometimes if the program dies, it's really annoying to not be able to
terminate a running trace. Add option -k to manually kill such a trace.

18 years ago[PATCH] Read and sort events in chunks from stdin
Jens Axboe [Mon, 5 Sep 2005 13:51:33 +0000 (15:51 +0200)]
[PATCH] Read and sort events in chunks from stdin

This should improve performance of the piped live tracing and at the same
time add support for sorting of incoming events at least in batches.

18 years ago[PATCH] Better strcmp fix
Nathan Scott [Mon, 5 Sep 2005 11:12:48 +0000 (13:12 +0200)]
[PATCH] Better strcmp fix

I looked at that strcmp memleak again for some reason, and wondered why not
use strcasecmp instead to further simplify things?

18 years ago[PATCH] Support for live tracing
Jens Axboe [Mon, 5 Sep 2005 11:09:45 +0000 (13:09 +0200)]
[PATCH] Support for live tracing

This patch adds support for doing live traces again, by allowing pipe of
data from blktrace to blkparse. A structural change was needed in blk_io_trace
to pass the CPU number info directly instead of inferring it from the filename.

To parse realtime:

$ blktrace -d /dev/xxx -o - | blkparse -i -

18 years ago[PATCH] Style fixups
Jens Axboe [Sun, 4 Sep 2005 15:37:06 +0000 (17:37 +0200)]
[PATCH] Style fixups

Alan seems to be fond of replacing tabs with space.s

18 years ago[PATCH] Stop threads before stopping trace
Jens Axboe [Sun, 4 Sep 2005 15:31:23 +0000 (17:31 +0200)]
[PATCH] Stop threads before stopping trace

Otherwise the BLKSTOPTRACE will be called with the relayfs fd still being
open, and the <dev> directory wont be removed then causing subsequent
traces to fail.

18 years ago[PATCH] Update kernel patch to match new relayfs layout
Jens Axboe [Fri, 2 Sep 2005 12:26:24 +0000 (14:26 +0200)]
[PATCH] Update kernel patch to match new relayfs layout

Lay out the relayfs files a little nicer, so it's not so confusing to look
at.

Original patch from Nathan Scott <nathans@sgi.com>

18 years ago[PATCH] Fix the stop trace logic to not attempt to stop it twice
Jens Axboe [Fri, 2 Sep 2005 11:47:23 +0000 (13:47 +0200)]
[PATCH] Fix the stop trace logic to not attempt to stop it twice

18 years ago[PATCH] blktrace fixes
Jens Axboe [Fri, 2 Sep 2005 11:17:03 +0000 (13:17 +0200)]
[PATCH] blktrace fixes

- Threads don't close the trace fd
- Switch to newer /relay/block/<dev>/trace<cpu> naming

18 years ago[PATCH] Kill silly changelog, it's too bothersome
Jens Axboe [Fri, 2 Sep 2005 08:57:07 +0000 (10:57 +0200)]
[PATCH] Kill silly changelog, it's too bothersome

18 years ago[PATCH] Fix usage of uninitialized array in pdu extraction
Nathan Scott [Fri, 2 Sep 2005 08:56:20 +0000 (10:56 +0200)]
[PATCH] Fix usage of uninitialized array in pdu extraction

18 years ago[PATCH] Fix memory leak in blktrace arg parsing
Nathan Scott [Fri, 2 Sep 2005 08:54:37 +0000 (10:54 +0200)]
[PATCH] Fix memory leak in blktrace arg parsing

18 years ago[PATCH] blktrace-fix-double-close
Nathan Scott [Fri, 2 Sep 2005 08:52:53 +0000 (10:52 +0200)]
[PATCH] blktrace-fix-double-close

stop_trace() has already closed the devfd, don't do so again.

18 years ago[PATCH] Remember to close devfd if start_trace() fails
Jens Axboe [Fri, 2 Sep 2005 08:51:59 +0000 (10:51 +0200)]
[PATCH] Remember to close devfd if start_trace() fails

18 years ago[PATCH] Make relayfs mount path an option
Nathan Scott [Fri, 2 Sep 2005 08:50:00 +0000 (10:50 +0200)]
[PATCH] Make relayfs mount path an option

Provide an option allowing blktrace to find relayfs at mount
points other than /relay (keeps /relay as the default though,
of course).

18 years ago[PATCH] Move BLK_TA_QUEUE trace
Jens Axboe [Thu, 1 Sep 2005 15:25:00 +0000 (17:25 +0200)]
[PATCH] Move BLK_TA_QUEUE trace

It makes more sense to do it after request allocation.

18 years ago[PATCH] Always make sure we stop the trace properly
Jens Axboe [Thu, 1 Sep 2005 14:43:36 +0000 (16:43 +0200)]
[PATCH] Always make sure we stop the trace properly

18 years ago[PATCH] Fix oops on starting trace on stacked devices (such as md, dm, etc)
Jens Axboe [Thu, 1 Sep 2005 14:43:04 +0000 (16:43 +0200)]
[PATCH] Fix oops on starting trace on stacked devices (such as md, dm, etc)

18 years ago[PATCH] Timestamp updates
Alan D. Brunelle [Thu, 1 Sep 2005 13:44:54 +0000 (15:44 +0200)]
[PATCH] Timestamp updates

- Changed output format to seconds.nanoseconds (makes it easier on the eyes)
- Made a single time stamp reference point (instead of a per-cpu reference
  point)

18 years ago[PATCH] Sync inter processor time
Jens Axboe [Thu, 1 Sep 2005 13:43:12 +0000 (15:43 +0200)]
[PATCH] Sync inter processor time

Calibrate the time offsets of each processor in an SMP system on boot time
to attempt to keep the timestamps offsets in check.

18 years ago[PATCH] Update kernel patch to 2.6.13-mm1
Jens Axboe [Thu, 1 Sep 2005 11:11:25 +0000 (13:11 +0200)]
[PATCH] Update kernel patch to 2.6.13-mm1

18 years ago[PATCH] Add license and copyright information
Jens Axboe [Wed, 31 Aug 2005 07:41:36 +0000 (09:41 +0200)]
[PATCH] Add license and copyright information

18 years ago[PATCH] Add argument parsing for blktrace
Alan D. Brunelle [Wed, 31 Aug 2005 06:49:05 +0000 (08:49 +0200)]
[PATCH] Add argument parsing for blktrace

This patch adds the option of specifying the trace mask for blktrace, so you
can register an interest in only certain events.

18 years ago[PATCH] blktrace: fix payload tracing
Jens Axboe [Mon, 29 Aug 2005 13:49:05 +0000 (15:49 +0200)]
[PATCH] blktrace: fix payload tracing

Endianness conversion was done before reading ->pdu_len, thus screwing it
up and hanging forever waiting for data.

18 years ago[PATCH] stats update
root [Mon, 29 Aug 2005 12:45:44 +0000 (14:45 +0200)]
[PATCH] stats update

Split stats into per-cpu and don't display empty stats

18 years ago[PATCH] Split output stats into per-cpu
Jens Axboe [Mon, 29 Aug 2005 12:00:50 +0000 (14:00 +0200)]
[PATCH] Split output stats into per-cpu

18 years ago[PATCH] blkparse: Don't clear rb_root in sort_entries()
Jens Axboe [Mon, 29 Aug 2005 11:27:35 +0000 (13:27 +0200)]
[PATCH] blkparse: Don't clear rb_root in sort_entries()

It will override the entries already sorted from the previous CPUs.

18 years ago[PATCH] Always store trace data in big endian format
Jens Axboe [Mon, 29 Aug 2005 07:45:41 +0000 (09:45 +0200)]
[PATCH] Always store trace data in big endian format

18 years ago[PATCH] Don't overwrite magic in trace
Jens Axboe [Mon, 29 Aug 2005 08:14:11 +0000 (10:14 +0200)]
[PATCH] Don't overwrite magic in trace

Leave the _dat file traces completely intact, get the CPU number from the
file in blkparse instead. Without this change, it's impossible for other
users to verify the sanity and version of a trace.

18 years ago - Improved error handling
Jens Axboe [Mon, 29 Aug 2005 06:45:19 +0000 (08:45 +0200)]
- Improved error handling
- Remove _dat files, include the payload in _out
- Support for payload > 64 bytes
- Add install target
- Fix RELAYFS Kconfig selection (kernel patch)

18 years agoChangelog
Jens Axboe [Sat, 27 Aug 2005 12:17:09 +0000 (14:17 +0200)]
Changelog

18 years agoVarious cleanups and killing unused variables
Jens Axboe [Sat, 27 Aug 2005 12:16:56 +0000 (14:16 +0200)]
Various cleanups and killing unused variables

18 years agoFixed the sorting and parsing of traces with payloads
Jens Axboe [Sat, 27 Aug 2005 07:38:00 +0000 (09:38 +0200)]
Fixed the sorting and parsing of traces with payloads

18 years agoblkparse doesn't take a NR_CPU argument anymore
Jens Axboe [Fri, 26 Aug 2005 12:38:26 +0000 (14:38 +0200)]
blkparse doesn't take a NR_CPU argument anymore

18 years agoAdd the required relayfs-read update patch as well
Jens Axboe [Fri, 26 Aug 2005 12:21:40 +0000 (14:21 +0200)]
Add the required relayfs-read update patch as well

18 years agoUpdate README and add kernel patch
Jens Axboe [Fri, 26 Aug 2005 12:19:57 +0000 (14:19 +0200)]
Update README and add kernel patch

18 years agoFix typo in last fix
Jens Axboe [Fri, 26 Aug 2005 12:13:28 +0000 (14:13 +0200)]
Fix typo in last fix

18 years agosprintf -> snprintf
Jens Axboe [Fri, 26 Aug 2005 12:12:09 +0000 (14:12 +0200)]
sprintf -> snprintf

18 years agoIncrease MAX_CPUS to 512. Use the full magic number for cpu number, not
Jens Axboe [Fri, 26 Aug 2005 12:06:49 +0000 (14:06 +0200)]
Increase MAX_CPUS to 512. Use the full magic number for cpu number, not
just the bottom 8 bits.

18 years agoInitial commit
Jens Axboe [Fri, 26 Aug 2005 11:49:39 +0000 (13:49 +0200)]
Initial commit