Commit Graph

1645 Commits

Author SHA1 Message Date
Andrew Johnson
1735a821db Update pvDatabase 2025-02-11 17:44:51 -06:00
9c1334ae15 silence Codacy warning 2025-02-10 08:47:59 +01:00
Michael Davidsaver
9481deacb0 fdManagerTest: quiet codacy warnings 2025-02-10 08:47:59 +01:00
8f1a3888c6 cleanup coding style in fdManager 2025-02-10 08:47:59 +01:00
c9183b5241 fdManager: filter poll flags for Window's sake 2025-02-10 08:47:59 +01:00
c3f57ee818 make fdManagerPrivate::fdReg volatile to avoid codacy warning 2025-02-10 08:47:59 +01:00
27f4261dfb use smart pointers in fdManager 2025-02-10 08:47:59 +01:00
5eb9997791 fix codacy warning: make fdManagerPrivate constructor explicit 2025-02-10 08:47:59 +01:00
bfc2f832ec fdManager uses poll() on Darwin too 2025-02-10 08:47:59 +01:00
f09b235fce Keep implementation details of fdManager out of header file 2025-02-10 08:47:59 +01:00
ece031c88b fdManager use std::vector 2025-02-10 08:47:59 +01:00
cbbbd67843 fdManager uses poll() on Windows and RTEMS too
RTEMS needs to use the "new" network stack
Windows has poll since Vista
Don't use poll on cygwin: it emulates poll() using select().
2025-02-10 08:47:59 +01:00
57c0295024 fdManager changed to use poll()
The implementation using select() limits file desciptors to FD_SETSIZE,
typically 1024 on Linux. This number is too low for some applications,
for example for the CA gateway.
Therefore, Linux builds use poll() instead.
2025-02-10 08:47:59 +01:00
Michael Davidsaver
8f77e941c7 add fdManager test 2025-02-10 08:47:59 +01:00
Michael Davidsaver
c76395abc6 remove fdmgrTest
Not a unittest, and not functional.
2025-02-10 08:47:59 +01:00
Michael Davidsaver
1d19ba4cc2 doc and log for *MustSucceed()
Replace the "never returns NULL." statement which is
manifestly not true.
2025-02-07 09:09:22 -08:00
JJL772
144f9756ea Add iocshSetError in a bunch of places 2025-02-05 10:16:16 -06:00
DW
48eed22f3b check IOCSH_STARTUP_SCRIPT before set 2025-02-05 10:03:35 -06:00
Érico Nogueira
8ac2c87156 Post monitors from compress record when resetting it
This way clients receive updates with an empty array after writing into
the RES field.
2025-02-05 09:59:18 -06:00
Timo Korhonen
065fe7cab6 Fix internal link 2025-01-28 15:05:55 -08:00
Michael Davidsaver
e4ad4becde rationalize osdMutex
Avoids split allocation.
Eliminates special case free-list.

win32: eliminate pre-XP
rtems-score: eliminate non-fast
2024-12-29 16:35:26 -08:00
Michael Davidsaver
1cd141c540 dbLoadTemplate: error propagation and more context on .db syntax error 2024-12-29 16:35:26 -08:00
72026a27a0 allow to load the same alias multiple times (unless dbRecordsOnceOnly is set) 2024-12-29 16:35:25 -08:00
Simon Rose
9fb820b46e Add tests for relative include paths 2024-12-29 16:34:10 -08:00
Jeremy Lorelli
721e9cc3a7 Add ABORT_ON_ASSERT flag to CONFIG_SITE
This flag causes EPICS to call abort() on assertion failures rather than
suspend the executing thread. With the epicsThreadSuspendSelf() behavior,
an IOC can end up in a difficult to detect error state where one or more
threads has essentially crashed due to an assertion failure.

This also matches the C behavior of assert(3)
2024-12-29 16:34:10 -08:00
Simon Rose
0186836449 Reading a file from the command-line should not use include paths 2024-12-29 16:34:10 -08:00
Simon Rose
b90ab7de13 Improve relative path check for msi
For posix-ish systems, we previously checked that a path was relative
by simpy checking if it included a '/' character. This meant that you
could not, for example, do
```
$ cat foo.substitutions
file rel/to/bar.template {
}

$ msi -I /some/path foo.substitutions
```
where our template file is located at `/some/path/rel/to/bar.template`.

Note that relateive paths work differently on Windows, so we carve out
an exception there.
2024-12-29 16:34:10 -08:00
Jeremy Lorelli
333be085c0 Fix compile error in tsSLList.h 2024-12-28 12:40:27 -08:00
Ralph Lange
5a11954c51 Add documentation for the UDFS field 2024-12-26 18:44:23 +01:00
Jure Varlec
d0cf47cd6f Propagate AMSG through MSS links
MS and MSI links do not propagate STAT and therefore do not propagate
AMSG, either. CA, CP and CPP links also do not propagate AMSG, but the
reason is technical: the message is not available over Channel Access.
2024-12-11 09:38:02 -06:00
dac620a708 loop safe wrapper for dbGet added 2024-12-11 09:31:49 -06:00
4ee766b6b1 correct mask for finding field in linkGrp
It worked before because the get_xxx functions are never called for
the links (bit0 = 1), but checking both bits looks cleaner.
2024-12-11 09:31:49 -06:00
5143258011 fix support for link 0 2024-12-11 09:31:49 -06:00
Michael Davidsaver
ead8b7e82b doc getIocState() 2024-12-09 17:00:03 -08:00
9b8712e095 Merge remote-tracking branch 'github/recursion_bug_v2' into PSI-7.0 2024-12-05 16:59:12 +01:00
d13441ebba Revert "fix recursion bug"
This reverts commit 7d73932dc8.
2024-12-05 16:58:32 +01:00
b53505d339 Merge branch '7.0' into PSI-7.0 2024-12-05 16:58:15 +01:00
75524e43bc loop safe wrapper for dbGet added 2024-12-02 11:18:42 +01:00
1567da1366 correct mask for finding field in linkGrp
It worked before because the get_xxx functions are never called for
the links (bit0 = 1), but checking both bits looks cleaner.
2024-11-29 10:47:23 +01:00
35ba9d65a2 fix support for link 0 2024-11-29 10:42:07 +01:00
Jure Varlec
8483ff9560 Clear NAMSG together with NSTAT and NSEV
Previously, there was no way to clear AMSG when setting a no-alarm
status: it kept the last error message, and it is strange to have a
record that has no alarm while showing an error. With this change, the
NAMSG field is treated the same as NSTA and NSEV: they are cleared when
their content is moved to STAT, SEVR, and AMSG.
2024-11-27 07:40:39 -08:00
7d73932dc8 fix recursion bug
Previously, linking a field of e.g. a calc record to itself
(calc.INPA = calc.A) directly or through a chain of other records
that support to follow the link for attributes like units, precision
caused an infinite recursion and crashed the ioc.

Affected records: calc, calcout, seq, sub and aSub

Also added support for seq link 0 fields which had been added meanwhile.
2024-11-25 11:32:41 +01:00
b7cc33c3c9 Send DBE_PROPERTY events before DBE_VALUE | DBE_LOG events to allow clients to update their formatting etc in time 2024-11-13 10:00:12 -06:00
9e7cd241e0 fix DBE_PROPERTY events for mbbi/mbbo records 2024-11-13 10:00:12 -06:00
faac1df1c6 send DBE_PROPERTY events only if property field actually changed 2024-11-13 10:00:12 -06:00
Chris Johns
594e5ee9db rtems: Set the name of EPICS threads so RTEMS commands report the name
Use the NP POSIX call pthread_setname_np to set the POSIX thread name.
2024-10-31 16:21:49 -05:00
Chris Johns
135193c4bd rtems: Fix shell networking commands
This patch cleans up the shell defines for the libbsd and legacy
network stack shell commands.

The top and rtems shell commands have been added.
2024-10-31 16:21:49 -05:00
Chris Johns
9d65da5c61 rtems: Fix rtems_bsdnet_bootp_server_address decl with RTEMS_LEGACY_STACK
The legacy stack provides this variable and this decl clashed. This was
seen before now due to the link order of libraries.
2024-10-31 16:21:49 -05:00
Chris Johns
83cbf8a4c7 freebsd: Add support for x86 and amd64 builds
Use the OS compiler, ie cc and c++.
2024-10-31 16:16:45 -05:00
Michael Davidsaver
f6cd43b7cc asTrap serverSpecific is dbChannel
Save/restore dbAddr::pfield around callbacks to
avoid corruption if CB forgets to restore.

Need to peak at dbChannel.h during libCom build.
So generate dbCoreAPI.h early, and add extra
-I to source location when compiling dbTrapWrite.c
2024-10-31 12:30:15 -05:00