diff options
Diffstat (limited to 'lib/libpthread/NOTES_OSR5_BUILD_SKUNKWARE97')
| -rw-r--r-- | lib/libpthread/NOTES_OSR5_BUILD_SKUNKWARE97 | 45 |
1 files changed, 45 insertions, 0 deletions
diff --git a/lib/libpthread/NOTES_OSR5_BUILD_SKUNKWARE97 b/lib/libpthread/NOTES_OSR5_BUILD_SKUNKWARE97 new file mode 100644 index 00000000000..fd2e00c9860 --- /dev/null +++ b/lib/libpthread/NOTES_OSR5_BUILD_SKUNKWARE97 @@ -0,0 +1,45 @@ +This port to Osr5 was donated by + ARTURO MONTES mitosys@colomsat.net.co +Its a snapshot of the 1.60 Beta 5 version + +It passes all the tests except test_fork (this seems to be a problem +with the MIT source) +and I suspect there may also to be problems with +the floating point initialisation and perhaps the netdb access as well + + +Regarding the networking API's mind the following (ARTURO): + +- Always we are searching host address in /etc/hosts by default, if you +want other behaviour let me to know. + +- If you haven't an entry in /etc/hosts, please create it and after test +with DNS service. + +regarding floating point initialisation (ARTURO) + +...mit-pthreads/pthreads/tests> ./test_preemption_float +test_preemption_float INDETERMINATE + +I inhibit the pthread float code, but they are a minor changes in machdep.c. + +> What did you do/needs to be done here ?? + +Look into machdep.c machdep_save_float_state() and +machdep_restore_float_state() routine. I think the initialization float +code can have problems with pthread initialization code. + +---------------- + + +If you wish to rebuild from source you'll need the gnus devsys +(gmake and gcc) and don't use configure to reconfigure the makefiles, etc + ( I don't know why this is but it doesn't work) + +If Arturo keeps me updated with buildable snapshots I'll endeavour to keep +the skunkware Website updated with them. + + +Best of luck + +hops 01-Aug-97 |