Commit Graph
3663 Commits
Author SHA1 Message Date
Keenan Lang dfe6515809 offset by a space 2014-08-21 16:12:11 -05:00
Keenan Lang 76bd45cb45 Reverted calls that didn't need to use macDefExpand to single parameter 2014-08-21 16:09:21 -05:00
Keenan Lang 6d3b70251c Returned macEnvExpand to original parameters, added macDefExpand 2014-08-21 16:05:55 -05:00
Keenan Lang 6d38cc1f76 Returned iocshCmd to it's original function signature, added iocshLoadFile instead 2014-08-21 15:19:50 -05:00
Keenan Lang 6ebbf0dd8c Whitespace changes snuck in due to editor 2014-08-20 16:58:50 -05:00
Keenan Lang 0cb495bdb3 Sidestepping testing in macEnvExpandTest 2014-08-19 16:44:15 -05:00
Keenan Lang 69c7f0b065 Changing macEnvExpand to accomodate macro definitions 2014-08-19 16:40:51 -05:00
Keenan Lang ab94bb46f2 Changing iocshCmd to idealized syntax 2014-08-19 16:34:41 -05:00
Andrew Johnson 9079c8aa53 Reverted changes to epicsTime.h and epicsThread.h 2014-08-18 22:55:07 -05:00
Andrew Johnson 5e0bf53fec epicsTypes.h: Don't use C99 types
The __STDC_VERSION__ >= 199901L test never succeeds
in C++, and there were issues with the epicsInt8
definition anyway (int8_t is signed, but char might
not be).

I also removed the epicsBoolean enum which we never use.
2014-08-15 12:24:31 -05:00
Andrew Johnson 547a9b3ebc Tweak some dependency rules for correctness 2014-08-05 14:03:37 -05:00
Andrew Johnson aedea0da37 default/osdSpin.c add missing #include "cantProceed.h"
Also adjusted some error messages for common style.
2014-08-02 23:15:43 -05:00
Andrew Johnson b7f8f17227 blockingSockTest: Undo member renames using m_ prefix
Using the m_ prefix to denote class data members is dangerous in
code that has to compile on VxWorks 5.5 (which 3.15 does) because
the net/mbuf.h header contains the following macro definitions:

#define m_next          mBlkHdr.mNext
#define m_len           mBlkHdr.mLen
#define m_data          mBlkHdr.mData
#define m_type          mBlkHdr.mType
#define m_flags         mBlkHdr.mFlags
#define m_nextpkt       mBlkHdr.mNextPkt
#define m_act           m_nextpkt
#define m_pkthdr        mBlkPktHdr
#define m_ext           pClBlk
#define m_extBuf        m_ext->clNode.pClBuf
#define m_extFreeRtn    m_ext->pClFreeRtn
#define m_extSize       m_ext->clSize
#define m_extRefCnt     m_ext->clRefCnt
#define m_extArg1       m_ext->clFreeArg1
#define m_extArg2       m_ext->clFreeArg2
#define m_extArg3       m_ext->clFreeArg3
#define  m_copy(m, o, l)        m_copym((m), (o), (l), M_DONTWAIT)
#define  m_copym(m, o, l, w)    netMblkChainDup(_pNetDpool, (m), (o), (l), (w))
#define  m_freem(m)             netMblkClChainFree(m)
#define  m_free(m)              netMblkClFree(m)    
#define m_mbufs         mNum
#define m_drops         mDrops
#define m_wait          mWait
#define m_drain         mDrain
#define m_mtypes        mTypes

While the names in the original change all appear to be safe, I
don't want us to start using this convention because we will then
have to police commits more closely or risk build failures in the
future when someone who doesn't know about or remember the issue
can't or doesn't try building their code against VxWorks 5.5.

I don't particularly like the this-> convention to mark member
names either, but until the VxWorks 5.5 support requirement goes
away the m_ convention is worse. Another common convention of
using a leading underscore is not safe either, those names are
reserved for the C/C++ implementation.
2014-08-02 14:02:43 -05:00
Andrew Johnson 8d5cdc3747 Fix typo in macro name
Not that it really matters since only VxWorks and RTEMS actually
compile this file and they don't have shared libraries...
2014-08-02 13:13:49 -05:00
Andrew Johnson f1a59aa64a Use epicsUInt types, stdint.h is not found on all platforms 2014-08-02 13:09:07 -05:00
Andrew Johnson 6767f5089f threadPool: Replaced errno.h status values with S_pool_* ones
ETIMEDOUT is not provided on MinGW.
2014-08-01 17:55:53 -05:00
Andrew Johnson 59a0fa9364 Fix vxWorks again, passes the tests on all my CPUs now. 2014-08-01 16:06:17 -05:00
Andrew Johnson fcd0a4c75d Fixes for VxWorks implementation. 2014-07-31 15:43:35 -05:00
Andrew Johnson ab78480d54 Merged spinlockfix branch.
Some last-minute fixes for VxWorks build warnings.
2014-07-31 12:45:40 -05:00
Andrew Johnson 68bcc77885 Merged alarm-strings branch. 2014-07-31 10:17:24 -05:00
Andrew Johnson b3afbea7cf Merged Michael's thread-pool branch.
Made various formatting changes, and fixed one problem in the
test program which failed a test on UP machines.
2014-07-29 16:05:24 -05:00
Michael Davidsaver 716f2679a7 thread pool: don't use reserved names
Avoid global symbols with leading underscore
2014-07-29 12:21:07 -04:00
Michael Davidsaver b1a8b2f20e thread pool: switch thread counts to unsigned int
size_t is considered overly optimistic
2014-07-29 12:18:19 -04:00
Michael Davidsaver 6fee83900e don't include errCommon.h
doesn't exist anymore
2014-07-29 12:06:24 -04:00
Andrew Johnson 75da9fd454 Moved dlopen version of osdFindSymbol.c into posix
It was identical in solaris, Darwin and Linux, and had
not been included for cygwin so this reduces duplication.
2014-07-28 12:27:02 -05:00
Michael Davidsaver c4a1208d6e epicsSpinTest: add to libCom test harness 2014-07-25 18:17:27 -04:00
Michael Davidsaver 9cb65e5408 epicsSpinTest: plug some leaks 2014-07-25 18:17:02 -04:00
Andrew Johnson 2a9d05248f Fix epicsSpinTest.c spinThread tests
Runs many more rounds, without blocking with the lock held.
2014-07-25 16:21:03 -05:00
Andrew Johnson 752549d1c8 Fix epicsAtomic headers when used from C code
Several C++ and C99-isms crept in.
2014-07-25 15:05:58 -05:00
Michael Davidsaver 8fec9d6e00 epicsSpin: try lock return non-blocking
Avoid cantProceed() in try lock, even for undefined behavior.
2014-07-25 15:12:02 -04:00
Michael Davidsaver 5824f98972 epicsSpinTest: fix verifyTryLock()
avoid sleeping with a spinlock held.
Now test only works on SMP systems.
2014-07-25 15:11:54 -04:00
Michael Davidsaver b5fe4abfa8 switch to use sysInWord and friends 2014-07-25 11:15:27 -04:00
Michael Davidsaver 95b916ecd4 thread pool: mark epicsJobCreate() as safe for job functions
Also, use epicsJobMove() to avoid some redundant code
2014-07-25 10:16:20 -04:00
Michael Davidsaver cbfbce54f9 thread pool: epicsThreadPoolWait return ETIMEOUT 2014-07-25 09:59:46 -04:00
Michael Davidsaver b3a1fe9c6b thread pool: fix return of epicsJobUnqueue()
Return 0 on success (was queued, now is not),
1 if not queued initially, and EINVAL
if orphaned.
2014-07-24 18:56:27 -04:00
Michael Davidsaver 83dfc7980d thread pool: handle failure to create worker
epicsJobQueue() returns EAGAIN when the first worker
can't be lazily created.

Failure to create workers beyond the first is
silently ignored.
2014-07-24 18:50:19 -04:00
Michael Davidsaver 69d1063067 thread pool: epicsJobQueue return EPERM
When pool control prevents operation
2014-07-24 18:19:33 -04:00
Andrew Johnson 9b6e270b97 Final spinlock tidying-up
* Abort epicsSpinTest() if epicsSpinCreate() returns NULL
* Adjust RELEASE_NOTES that describe the implementations.
2014-07-24 16:33:35 -05:00
Michael Davidsaver fc4119094f thread pool: epicsThreadPoolGetShared accepts NULL as defaults 2014-07-24 14:44:28 -04:00
Michael Davidsaver 01a50b5165 thread pool test 2014-07-24 14:22:52 -04:00
Michael Davidsaver 658bd0b570 thread pool implementation 2014-07-24 14:22:52 -04:00
Michael Davidsaver a8fd14aae1 Thread pool API 2014-07-24 14:22:52 -04:00
Michael Davidsaver 87a6688c17 epicsSpin: remove redundant cantProceed() messages 2014-07-24 10:23:55 -04:00
Michael Davidsaver 3ba5bf943e epicsSpin: better error messages when mis-use is detected 2014-07-23 18:40:08 -04:00
Andrew Johnson 6bdb8f911a Suppress errlog output of expected warning messages. 2014-07-23 16:18:32 -05:00
Michael Davidsaver 0bf17be9e3 mmio: vxWorks m68k use default, add weak symbols for old versions 2014-07-16 15:12:56 -04:00
Andrew Johnson 2fb0b0763d Implement 64-bit database field types.
Incorporates major reworking of the dbConvert.c file.
if the 64-bit stuff gets rejected most of that should
get pulled in anyway.
2014-07-10 18:13:28 -05:00
Andrew Johnson 3ff49fc9b2 Fixed build errors, remove dependency on std.
Build errors: Include guard missing from epicsUnitTest.h;
    dbShutdownTest.c was calling the non-universal strcasecmp()
    instead of epicsStrCaseCmp().

ioc/db/test no longer depends on std. I modified
    xRecord to make it a working record type, and simplified the
    other test programs so they all use the same new expanded
    dbd file rather than each making their own. I also added
    dbShutdownTest() to epicsRunDbTests().
2014-07-10 14:27:44 -05:00
Andrew JohnsonandMichael Davidsaver 60a0c7f181 Make epicsExit subsystem reusable.
Calling epicsAtExit() after epicsExitCallAtExits() now
recreates the per-process list and registers the routine.
2014-06-23 16:28:25 -04:00
Michael DavidsaverandMichael Davidsaver 3ca42fc838 cleanup iocsh 2014-06-23 16:28:22 -04:00