On 5/12/21 12:31 AM, Richard Biener wrote:
Otherwise looks OK. Did you check for write_symbols uses in FEs and targets?
Richard.
Yes, I have. I must admit I have gone back and forth in my mind on this.
My initial thinking was to adjust only those checks where I expect more
than 1 write_symbols. Because in most contexts, e.g., (write_symbols ==
XCOFF_DEBUG) is a stricter check than (write_symbols & XCOFF_DEBUG), in
some sense.
Anyway, after you raised this point, however, I looked again, and have
added some dwarf_debuginfo_p usages in the second patch. Basically, I
replaced some write_symbols == DWARF2_DEBUG with dwarf_debuginfo_p ()
usage. Other than DWARF2_DEBUG, at this time, no other debug format is
expected to be written out together with other debug formats.
As for other usages of write_symbols
- I plan to get rid of the VMS_AND_DWARF2_DEBUG symbol in a subsequent
patch (this should deal with most usages in vmsdbgout.c).
I will post V2 next.
A question though - I have regression tested the V2 patch set on x86_64.
Although the sort of changes in the backend files can be OK, I am happy
to take any further suggestions for testing this patch set.
If you touch backends in non-trivial ways it's at least good to test it
still builds in a simple cross configuration which means
configuring for one of the supported target triplets and doing
make all-gcc (that only builds host objects so you don't need
binutils or target headers for this)
Richard.
Thanks. I tested the patch set for select triplets containing
c6x/mips/powerpc and cygwin/darwin. Found and fixed the build issues.
Will be posting V3 next.
Thanks
Indu
