Commit Graph

12998 Commits

Author SHA1 Message Date
Andrew Johnson
5b860ec3ec libCom: Use SIG_IGN to ignore signals.
Simplified the code in osdSignal, also solves problems if a handler is
already installed that uses the 3-argument callback.
2010-09-23 15:09:03 -05:00
Andrew Johnson
eacb6602bc libCom: Use OS NAN and INFINITY macros where available.
bug: 545385
2010-09-23 14:18:03 -05:00
Andrew Johnson
4e99cfc763 Added target support for Apple's iOS devices.
Tom Pelaia II and Mark Engbretson contributed the build configuration
and OS-dependent files, I modified the build config to bring it up to
R3.14.12 standards and added the Release Notes.
2010-09-22 16:26:15 -05:00
Andrew Johnson
a9b5f0c6a0 Merged Michael Abbott's dynamic-array branch, rebased.
I added Perl support, and wrote some release notes.
2010-09-20 16:21:50 -05:00
Andrew Johnson
dea1b9df3b Tweaked the release notes to describe which areas were not changed. 2010-09-20 14:41:33 -05:00
Andrew Johnson
7ce4eb96ea cap5: Support dynamic array sizes through the Perl API. 2010-09-17 14:11:00 -05:00
Andrew Johnson
3950153451 cap5: minor documentation updates/fixes. 2010-09-17 13:51:31 -05:00
Andrew Johnson
3b4eedaa26 cap5: caget should print limits using %g, not %d. 2010-09-17 12:49:22 -05:00
Andrew Johnson
6923f82119 dbtools: Modified and added test files. 2010-09-08 17:02:23 -05:00
Andrew Johnson
a7d7336392 dbtools: Added dbltExpand.c
This is a test program which implements the template expansion
capabilities of msi, but using the dbLoadTemplate.c code.  It should
be useful for testing, to make sure that both versions generate the
same output.  It won't work on Windows at the moment due to a clash in
the function decorations of the dbLoadRecords() function.
2010-09-08 16:59:10 -05:00
Andrew Johnson
46e723575e dbtools: This is Ben's version of msi.c that implements global {} and
restores the variable scope for every instantiation of the teplate. 
He's now having second thoughts on the design though, so this might
be not useful.
2010-09-08 16:56:13 -05:00
Janet Anderson
d4a9958002 Changes to allow rebuild to work with parallel builds option, -j. 2010-09-03 10:07:17 -05:00
Michael Davidsaver
aee66600ef errPrintf() newline fix.
errPrintf() was adding a newline when writing to the err log buffer,
but not when printing to the screen.  This results in the error
message running together with the next line.  Added a newline to the
stderr output.
2010-09-01 14:56:51 -05:00
Janet Anderson
6a0a263380 Applied PATH_FILTER to input libraries on linker command line
Now when using MSVC complier with non cygwin makes, RELEASE files can
contain directories starting with forward slashes.

bug:627511
2010-08-31 11:33:37 -05:00
Janet Anderson
13e7b1b0b7 Added HDEPENDS_CFLAGS to CFLAGS to create *.d dependancy files.
bug: 626859
2010-08-30 09:58:24 -05:00
Janet Anderson
5094762012 Added target list to install library rule, needed for make3.8.2 builds
bug:626848
2010-08-30 09:30:11 -05:00
Ralph Lange
866f062d0a toolsComm: Remove #ifdef checks for old VMS systems (fixes lp:625034) 2010-08-27 10:12:20 -04:00
Andrew Johnson
177a9f89c8 dbtools: Merged Ralph's msi-join branch, adds msi to Base. 2010-08-25 16:56:22 -05:00
Andrew Johnson
ac32658bf4 dbtools: Added another possibility for pattern_substitutions rule. 2010-08-25 13:21:24 -05:00
Andrew Johnson
e2464c8572 Implemented "global" variables in substitution files.
Rewrote the YACC grammar, with help from Benjamin Franksen in formally
defining the substitution file format in EBNF.  This version also
changes the way in which variables are added to the sub_collect
string; previously it appended a ',' after every entry, and removed it
off the end before calling dbLoadRecords(), but now we put the ',' at
the beginning of each entry, and just offer dbLoadRecords() the string
starting at the second character (not that this really matters, macLib
will quite happily ignore either a leading or a trailing comma in the
variable definition string).

We now warn if there are substitution values for which we have no
name, or if the file uses the deprecated syntax which permitted a
bareword token in front of the variable substitution or pattern
definition braces.
2010-08-24 19:01:19 -05:00
Ralph Lange
a0a9ee4553 ca: Avoid possible race condition during cac shutdown.
While the cac dtor runs and waits for TCP circuit threads to exit,
new circuits could be opened by name resolution replies on existing
(still active) circuits.
Avoid this by setting a flag (lock being held) in the dtor, and
checking it in cac::transferChanToVirtCircuit.
2010-08-23 16:17:16 -04:00
Ralph Lange
6b6de69f1d build: Point default MSI to the base version. 2010-08-23 15:08:24 -04:00
Ralph Lange
52f4c8ae8f dbtools: shuffle files towards a reasonable layout, keep test files in test dir 2010-08-23 14:48:43 -04:00
Ralph Lange
2a19c5f0b8 dbtools: Join msi repository into ../src/dbtools 2010-08-23 14:36:23 -04:00
Andrew Johnson
4a637c04cf dbtools: Added command-line variable parameters. 2010-08-20 17:59:38 -05:00
Andrew Johnson
3db5427e1d dbtools: Source code clean-up, code layout changes only. 2010-08-20 17:50:52 -05:00
Andrew Johnson
90ab284583 db: Replace forward reference to notifyCallback, got lost in rebase. 2010-08-20 00:40:26 -05:00
Andrew Johnson
85c2877746 Document the dynamic-array changes. 2010-08-17 18:06:19 -05:00
Andrew Johnson
a6a01e6822 Converted buckTest.c to use epicsTime.h instead of time.h
This should finally resolve the build conflict between RTEMS-uc5282
and Windows, and remove the build warning on vxWorks.
2010-08-17 17:53:43 -05:00
Jeff Hill
7af0397d58 Examining the previous change I see that it will introduce a race condition when circuits are being created and destroyed at close to the same instant in time. See revision 11786 of cac.cpp which fixes mantis 334 if you are interested in what this code does. After running the regression tests, I pushed in a fix (which is removing above mentioned change and restoring the increment of iiuExistenceCount in cac::findOrCreateVirtCircuit. 2010-08-16 17:45:06 -06:00
Jeff Hill
a3dd3e1a2f unistd.h isnt available on windows, and doesnt appear to be required by this code? 2010-08-16 17:34:39 -06:00
Andrew Johnson
09c0c43865 Cleanup a couple of compiler warnings. 2010-08-16 16:15:26 -05:00
Ralph Lange
2666e9d62b Bug fixes to CA client introduced by CA-over-TCP code. 2010-08-16 15:49:52 -05:00
Jeff Hill
1eb5546eb7 meged andrew's change 2010-08-13 18:31:00 -06:00
Jeff Hill
c1bd11de58 fixed bug where win32 version of epicsThreadHighestPriorityLevelBelow didnt map to a lower priority successfully if the epics priority was mapping into the 2nd win32 priority bucket 2010-08-13 18:15:49 -06:00
Andrew Johnson
25d1b89de1 Use $(RTEMS_VERSION) in default path to RTEMS_BASE. 2010-08-13 17:56:27 -05:00
Andrew Johnson
1bab392da7 Deleted trailing ';' from epicsPlacementDeleteOperator macro
(which looks like a member function, but isn't).
2010-08-13 13:12:06 -05:00
Andrew Johnson
0c4746dd66 Moved new variable definitions to the beginning of their block.
Older C compilers such as vxWorks 5.x insist on this.
2010-08-13 12:59:50 -05:00
Michael Abbott
053ab01db4 Guard memset in camessage.c to prevent wild memory overrun if mismatch
between payload and data size.
2010-08-11 17:37:56 -05:00
Michael Abbott
0291900562 Add CA message header size optimisation.
If a packet is sufficiently small it can be worth the small
optimisation of saving the 8 bytes in a large packet header.
2010-08-11 17:37:30 -05:00
Michael Abbott
b2374f583b Drive-by refactoring in caserverio.c
Minor refactoring of code block in cas_copy_in_header() to gather
common header code.
2010-08-11 17:36:58 -05:00
Michael Abbott
89ebfc6e9b Add autosizing array support for caget and camonitor. 2010-08-11 17:36:36 -05:00
Michael Abbott
519c94a6e3 Update rsrv to support V4.13 protocol extension.
The rsrv server is updated in this commit to deliver variable sized
waveforms in response to a zero length request, implementing an
extension to the v4 EPICS protocol tied to minor version 13.

The core of the changes are in read_reply, rsrv/camessage.c, which first
reserves packet space for the largest possible response, but then resizes
the packet according to how much data is actually returned from the
database.
2010-08-11 17:36:11 -05:00
Michael Abbott
ee4a615caa Client side support for automatic resizing arrays.
In this commit the channel access client is modified to allow a zero
size request in a caget or camonitor call.  This is passed through to
the corresponding CA_PROTO_READ_NOTIFY or CA_PROTO_EVENT__ADD message,
but only if the minor version of the protocol is >=13.

This commit also adds a new protocol test, CA_V413(), which detects a
server which claims to understand the new zero length request.

Finally, this commit prepares for a subtle change in the CA protocol.
A request for a zero length subscription or data request, which will
only be made for protocol version 4.13 and above, is interpreted as a
request for autosized data, where the intrinsic dynamic length of the
requested waveform should be returned.

With the synchronous ca_array_get() call there is no mechanism available
to pass back the size of the retrieved data if it is different from
the requested count, so the simplest thing is to ensure we don't
request autosizing data through this api.
2010-08-11 17:35:23 -05:00
Michael Abbott
e54b631b8d Guard memset in camessage.c to prevent wild memory overrun if mismatch
between payload and data size.
2010-08-11 16:49:46 -05:00
Michael Abbott
9d9ba1a442 Add CA message header size optimisation.
If a packet is sufficiently small it can be worth the small
optimisation of saving the 8 bytes in a large packet header.
2010-08-11 16:42:44 -05:00
Michael Abbott
30f0871c2b Drive-by refactoring in caserverio.c
Minor refactoring of code block in cas_copy_in_header() to gather
common header code.
2010-08-11 16:41:15 -05:00
Michael Abbott
aa23ab63f1 Add autosizing array support for caget and camonitor. 2010-08-11 16:39:52 -05:00
Michael Abbott
e87bbef75c Update rsrv to support V4.13 protocol extension.
The rsrv server is updated in this commit to deliver variable sized
waveforms in response to a zero length request, implementing an
extension to the v4 EPICS protocol tied to minor version 13.

The core of the changes are in read_reply, rsrv/camessage.c, which first
reserves packet space for the largest possible response, but then resizes
the packet according to how much data is actually returned from the
database.
2010-08-11 16:38:49 -05:00
Michael Abbott
a0c0c55347 Client side support for automatic resizing arrays.
In this commit the channel access client is modified to allow a zero
size request in a caget or camonitor call.  This is passed through to
the corresponding CA_PROTO_READ_NOTIFY or CA_PROTO_EVENT__ADD message,
but only if the minor version of the protocol is >=13.

This commit also adds a new protocol test, CA_V413(), which detects a
server which claims to understand the new zero length request.

Finally, this commit prepares for a subtle change in the CA protocol.
A request for a zero length subscription or data request, which will
only be made for protocol version 4.13 and above, is interpreted as a
request for autosized data, where the intrinsic dynamic length of the
requested waveform should be returned.

With the synchronous ca_array_get() call there is no mechanism available
to pass back the size of the retrieved data if it is different from
the requested count, so the simplest thing is to ensure we don't
request autosizing data through this api.
2010-08-11 16:22:54 -05:00