mirrors/emacs
1
Fork 0
mirror of git://git.sv.gnu.org/emacs.git synced 2025-12-25 06:50:46 -08:00
Code Issues Projects Releases Wiki Activity
4440 commits 242 branches 194 tags 7.7 GiB
Commit graph

918 commits

Author SHA1 Message Date
Gareth Rees
e12d5890e3 Improvements to the mps build/installation process: 
1. Build eventtxt on Mac OS X (add "eventtxt" as dependency for "all" target).
2. In configure, check for the existence of the header "sqlite3.h" and set EXTRA_TARGETS accordingly.
3. In top-level Makefile, install all targets in EXTRA_TARGETS.

Copied from Perforce
 Change: 180907
 ServerID: perforce.ravenbrook.com
 2013-02-05 13:34:42 +00:00
Gareth Rees
51b123525e Avoid error "variable 'res' set but not used" from gcc. 
Copied from Perforce
 Change: 180905
 ServerID: perforce.ravenbrook.com
 2013-02-04 17:32:13 +00:00
Gareth Rees
c6aeb6e62f Merge changes 180885 and 180887 from branch/2012-10-10/measurement to branch/2012-10-09/user-guide. 
Copied from Perforce
 Change: 180904
 ServerID: perforce.ravenbrook.com
 2013-02-04 16:44:34 +00:00
Gareth Rees
afd7af371b Don't #error if defined(mps_build_ll) and __has_builtin(__builtin_readcyclecounter). put a todo here instead, referencing job003411. 
Copied from Perforce
 Change: 180903
 ServerID: perforce.ravenbrook.com
 2013-02-04 16:40:28 +00:00
Nick Barnes
7dc00c2912 Get eventcnv/txt/sql working on fri3gc. 
Copied from Perforce
 Change: 180887
 ServerID: perforce.ravenbrook.com
 2013-01-31 14:54:35 +00:00
Nick Barnes
164f0a8a92 Port eventsql and friends to windows 7. 
Copied from Perforce
 Change: 180885
 ServerID: perforce.ravenbrook.com
 2013-01-30 16:06:59 +00:00
Gareth Rees
01b0ed6d11 Integrate changes from branch/2012-10-10/measurement. 
Copied from Perforce
 Change: 180506
 ServerID: perforce.ravenbrook.com
 2012-11-20 10:50:12 +00:00
Gareth Rees
0dd9586930 New type mps_label_t for telemetry labels. 
Deprecate "candy machine" interface to telemetry control.
Implement and document straightforward get/set/reset interface.
Put a TODO referencing job003387 and job003388.

Copied from Perforce
 Change: 180495
 ServerID: perforce.ravenbrook.com
 2012-11-16 16:42:41 +00:00
Gareth Rees
35e0ac259a New file "mpscmv2.h" just #include "mpscmvt.h", for backwards compatibility. 
Copied from Perforce
 Change: 180490
 ServerID: perforce.ravenbrook.com
 2012-11-16 14:48:53 +00:00
Gareth Rees
fcd7180d08 Rename "mpscmv2.h" to "mpscmvt.h" for consistency. 
Copied from Perforce
 Change: 180489
 ServerID: perforce.ravenbrook.com
 2012-11-16 14:48:14 +00:00
Nick Barnes
ba28c7a10d Exit(exit_success) after usage message. 
Copied from Perforce
 Change: 180484
 ServerID: perforce.ravenbrook.com
 2012-11-16 10:22:29 +00:00
Nick Barnes
151c6a1e63 Somehow i introduced an uninitialized variable bug into eventflush(). oops. 
Copied from Perforce
 Change: 180482
 ServerID: perforce.ravenbrook.com
 2012-11-15 22:07:28 +00:00
Nick Barnes
2df242c567 Remove assertion which requires a higher-resolution clock() than freebsd has. document an alternative mps_clock() implementation using getrusage(). 
Copied from Perforce
 Change: 180481
 ServerID: perforce.ravenbrook.com
 2012-11-15 22:06:49 +00:00
Nick Barnes
5a2950e2a1 Correct comment regarding mps_clocks_per_sec(). 
Copied from Perforce
 Change: 180480
 ServerID: perforce.ravenbrook.com
 2012-11-15 19:30:47 +00:00
Gareth Rees
490ee8a31c New event "arenasetemergency". 
Copied from Perforce
 Change: 180479
 ServerID: perforce.ravenbrook.com
 2012-11-15 14:08:13 +00:00
Gareth Rees
7043c844ae Restore "res = resok" to eventflush (removed in change 180152). 
Copied from Perforce
 Change: 180471
 ServerID: perforce.ravenbrook.com
 2012-11-14 17:46:03 +00:00
Gareth Rees
66f0e08954 Integrate nb's changes from the measurement branch. 
Copied from Perforce
 Change: 180465
 ServerID: perforce.ravenbrook.com
 2012-11-14 14:09:49 +00:00
Gareth Rees
6abf20535f Events have to be inside the arena lock. 
Copied from Perforce
 Change: 180463
 ServerID: perforce.ravenbrook.com
 2012-11-13 17:59:41 +00:00
Nick Barnes
686ee38101 Tidy up the event pipeline. 
Copied from Perforce
 Change: 180460
 ServerID: perforce.ravenbrook.com
 2012-11-13 12:51:58 +00:00
Gareth Rees
6f81d69df3 Add two new events: 
* ArenaAccess for handling barrier hits;
* ArenaPoll for choosing a quantum of work.

Copied from Perforce
 Change: 180432
 ServerID: perforce.ravenbrook.com
 2012-11-09 15:19:35 +00:00
Gareth Rees
2350b83b6b Use a typedef for the type of the function passed to mps_amc_apply to make the documentation clearer while remaining backwards-compatible. 
(This was change 179975 but got reversed accidentally by change 180236.)

Copied from Perforce
 Change: 180431
 ServerID: perforce.ravenbrook.com
 2012-11-09 11:15:16 +00:00
Gareth Rees
7645b348a4 Amcnailrangeisunmarked needs to take arguments as client pointers and look at the nails for the corresponding base pointers. 
Copied from Perforce
 Change: 180390
 ServerID: perforce.ravenbrook.com
 2012-11-07 17:17:05 +00:00
Gareth Rees
6b89b020c5 Add public interface for mfs pool class. 
Correct the documentation for MFS: constructor takes extend_by and unit_size in that order.

Copied from Perforce
 Change: 180372
 ServerID: perforce.ravenbrook.com
 2012-11-06 17:08:58 +00:00
Gareth Rees
d743f8b510 Support ambiguous interior pointers in amc by checking to see if any grain in an object is nailed (not just the first grain). 
AMCSegDescribe shows which grains are nailed.

Copied from Perforce
 Change: 180370
 ServerID: perforce.ravenbrook.com
 2012-11-06 15:47:21 +00:00
Gareth Rees
99bca8b5e2 Amc can't omit the guard "segrankset(seg) != ranksetempty" for calling segsetgrey, because amcz uses segments with an empty rank set. 
It's OK to call SegSetGrey on a segment with an empty rank set so long as you are not actually setting it to be grey for any traces, so correct the AVER accordingly.

Copied from Perforce
 Change: 180236
 ServerID: perforce.ravenbrook.com
 2012-11-01 17:21:57 +00:00
Nick Barnes
1330a1a99c Remove unused mps_t_longest type. 
Copied from Perforce
 Change: 180169
 ServerID: perforce.ravenbrook.com
 2012-10-30 17:14:45 +00:00
Nick Barnes
ecd77595bb Add mps_clocks_per_sec to the eventinit header, and put some header-checking code into eventcnv, so now we have at least a rudimentary check that a binary event log is readable. 
Copied from Perforce
 Change: 180160
 ServerID: perforce.ravenbrook.com
 2012-10-30 14:15:27 +00:00
Gareth Rees
fb1601827a Integrate nb's clock changes. mps_clock_t now an alias for mps_word_t. 
Copied from Perforce
 Change: 180153
 ServerID: perforce.ravenbrook.com
 2012-10-30 11:12:38 +00:00
Nick Barnes
1746c7583a Make eventclocksync events, and output them once per buffer flush. 
Copied from Perforce
 Change: 180152
 ServerID: perforce.ravenbrook.com
 2012-10-30 09:36:30 +00:00
Nick Barnes
271b4f6e0a change mps_clock_t to word. 
Copied from Perforce
 Change: 180151
 ServerID: perforce.ravenbrook.com
 2012-10-30 08:43:39 +00:00
Nick Barnes
7616ea7159 Take out debugging printfs. oops. 
Copied from Perforce
 Change: 180110
 ServerID: perforce.ravenbrook.com
 2012-10-26 19:16:59 +01:00
Richard Brooksby
5ee7e6dd84 Tidying up case-insensitive mps_telemetry_control implementation a bit. 
Copied from Perforce
 Change: 180109
 ServerID: perforce.ravenbrook.com
 2012-10-26 17:04:30 +01:00
Gareth Rees
cec0b453ee Merge nb's eventcnv changes and rewrite guide/debug and topic/telemetry accordingly. 
Copied from Perforce
 Change: 180103
 ServerID: perforce.ravenbrook.com
 2012-10-26 14:35:10 +01:00
Nick Barnes
d48d96f15e Add an eventinit event, giving various parameters including the event system version. 
Copied from Perforce
 Change: 180100
 ServerID: perforce.ravenbrook.com
 2012-10-26 13:22:04 +01:00
Nick Barnes
20ad19da97 Make mps_telemetry_control case-insensitive, and make it accept "all" as an event class. 
Copied from Perforce
 Change: 180098
 ServerID: perforce.ravenbrook.com
 2012-10-26 13:03:44 +01:00
Gareth Rees
6c6d7a00d6 No need to include "mpsavm.h". the comment justifying it was bogus: "only for mps_space_create". 
Copied from Perforce
 Change: 180093
 ServerID: perforce.ravenbrook.com
 2012-10-26 09:51:12 +01:00
Nick Barnes
750d515233 Further simplify the eventcnv text output format. 
Copied from Perforce
 Change: 180054
 ServerID: perforce.ravenbrook.com
 2012-10-24 16:10:46 +01:00
Nick Barnes
ec8692c49c Integrate gdr's somewhat cut-down eventcnv code. 
Copied from Perforce
 Change: 180046
 ServerID: perforce.ravenbrook.com
 2012-10-24 11:12:17 +01:00
Gareth Rees
9d22cb3c49 Remove comment following discussion with rb: "assumes pun compatibility between rank and mps_rank_t which is checked by mpsi_check in <code/mpsi.c>" as this is no longer true: ranks no longer appear in the public interface, so their pun compatibility is no longer checked. 
Copied from Perforce
 Change: 180031
 ServerID: perforce.ravenbrook.com
 2012-10-23 15:42:17 +01:00
Gareth Rees
07d0aaf113 Remove reporting features from eventcnv, as discussed with rb. eventcnv was always intended to be a simple tool that just translates events, not a report generator. in particular: 
* Remove -e option and associated event specification parsing.
* Remove -b option and bucket statistics.
* Don't intern labels (always print addresses).
* Remove -v option (events are always output).
* If -h or -? is specified, exit after printing help message.
* Help text refer to the "Telemetry" section of reference manual.

Copied from Perforce
 Change: 180028
 ServerID: perforce.ravenbrook.com
 2012-10-23 14:28:22 +01:00
Nick Barnes
1822b23637 Better table existence test. 
Copied from Perforce
 Change: 179985
 ServerID: perforce.ravenbrook.com
 2012-10-20 20:34:30 +01:00
Gareth Rees
88bcefa962 Use a typedef for the type of the function passed to mps_amc_apply to make the documentation clearer while remaining backwards-compatible. 
Copied from Perforce
 Change: 179975
 ServerID: perforce.ravenbrook.com
 2012-10-19 19:34:05 +01:00
Gareth Rees
3bada6378b Based on discussion with rb, reverse the sense of the comments on mps_ap_fill and mps_ap_trip. 
Copied from Perforce
 Change: 179971
 ServerID: perforce.ravenbrook.com
 2012-10-19 16:47:49 +01:00
Nick Barnes
f0757a1640 Add event_param glue table. 
Copied from Perforce
 Change: 179953
 ServerID: perforce.ravenbrook.com
 2012-10-19 07:32:40 +01:00
Nick Barnes
efdc6b4018 Re-jigged eventsql so it uses plain-text intermediate files and handles stdin ok. 
Copied from Perforce
 Change: 179947
 ServerID: perforce.ravenbrook.com
 2012-10-18 17:00:31 +01:00
Gareth Rees
373f8a9b54 Mps_shift_t is not used in the external interface, so following discussion with rb, remove it. 
Copied from Perforce
 Change: 179944
 ServerID: perforce.ravenbrook.com
 2012-10-18 16:23:59 +01:00
Nick Barnes
2dd20913e4 Added long explanatory comment, and license, to eventsql.c. removed eventsql.h (what was i thinking?). 
Copied from Perforce
 Change: 179915
 ServerID: perforce.ravenbrook.com
 2012-10-17 14:49:00 +01:00
Nick Barnes
fa8cda3e9c Add command-line controls to eventsql. 
Copied from Perforce
 Change: 179908
 ServerID: perforce.ravenbrook.com
 2012-10-17 03:03:54 +01:00
Nick Barnes
a47d640fe8 Eventsql goes hundreds of times faster (because all event inserts are now in a single transaction). 
Copied from Perforce
 Change: 179904
 ServerID: perforce.ravenbrook.com
 2012-10-16 22:19:54 +01:00
Nick Barnes
82a32fa22a Eventcnv was printing event clock values incorrectly: the upper 32 bits and then all the bits. have to mask the top bits when printing the bottom ones. 
Copied from Perforce
 Change: 179881
 ServerID: perforce.ravenbrook.com
 2012-10-15 00:35:40 +01:00