Opened 18 years ago
Last modified 11 years ago
#156 assigned defect
emxomfld: wlink hll support
Reported by: | bird | Owned by: | bird |
---|---|---|---|
Priority: | normal | Milestone: | |
Component: | emx | Version: | 0.6.5 |
Severity: | normal | Keywords: | wlink |
Cc: |
Description
This is mostly done, but need some finishing touches before the OpenWatcom guys will accept it I guess. I'll upload a patch some day so it won't get lost.
The initial wlink changes was done in #89.
Change History (8)
comment:1 by , 18 years ago
Milestone: | libc-0.6.3 → libc-0.6.4 |
---|
comment:2 by , 14 years ago
Status: | new → assigned |
---|
comment:3 by , 13 years ago
Milestone: | libc-0.6.5 → libc-0.6.6 |
---|
comment:4 by , 11 years ago
comment:5 by , 11 years ago
Milestone: | libc-0.6.6 |
---|
comment:6 by , 11 years ago
Component: | emx → libc |
---|---|
Milestone: | → libc-0.6.6 |
Version: | 0.6 → 0.6.5 |
Steven, could you please comment on that? Is there anything else that needs to be done before submitting patches?
A proper handler for the case when there is not enough memory that cleanly aborts WLINK with the appropriate message (instead of letting it hang for many hours) is perhaps one thing.
comment:7 by , 11 years ago
Component: | libc → emx |
---|---|
Milestone: | libc-0.6.6 |
comment:8 by , 11 years ago
The changes have all been submitted to the Perforce repository.
The handler should be in place. See
bld\wl\c\mem.c:198
in the OpenWatcom sources. The allocation policy allocates from low memory until there is less than 64MB left and then switches to allocating from upper memory. If the linker is going to run out of memory, it will run out attempting to allocate the buffer that holds the HLL_MISC_SECT section (i.e. the debug data) and it will have plenty of memory available for the cleanup to finish.
Cleanup time is the same whether or not the linker runs to completion without errors.
The extended run time you saw was most likely the result of building firefox with a wlink debug build. A wlink debug build runs additional code during cleanup and this code will run for hours whether or not the link finishes with errors.
There is some work remaining. I need to finish updating the OpenWatcom docs. Also, Knut's code indicates that some HLL debug data features are not yet implemented.
see also http://svn.netlabs.org/rpm/ticket/64