[PATCH] Slap a band-aid on PR41404 to restore bootstrap pending fuller fix.

[Dave Korn]

    Hi all,

  Following discussion in the audit trail, Honza has a plan to fix the
underlying issue in PR41404, which is a mishandling of const_string RTL when
generating the dwarf for the location tables in mem_loc_descriptor().

  That will take a little time to develop and review, so in the meantime this
patch shoves a temporary band-aid on the problem and restores bootstrap by
simply punting to track string values.  When Honza's patch arrives it will
implicitly revert this patch by re-adding something correct to replace the
deleted code in the CONST_STRING switch case.

gcc/ChangeLog:

	PR bootstrap/41404
	* dwarf2out.c (mem_loc_descriptor): Punt on CONST_STRING until
	we can handle it correctly.

  Fixes the PR's attached testcase and restores java bootstrap on
i686-pc-cygwin.  C tests running now, have only got as far as
gcc.c-torture/execute/20020328-1.c yet but no sign of any failures.  Y'all
know how lumbering Cygwin can be(*), it'll be way late tonight before it
completes, so would it be OK after say just C and Java testing?

    cheers,
      DaveK
-- 
(*) - our mascot isn't a hippo for nothing!

-------------- next part --------------
A non-text attachment was scrubbed...
Name: pr41404-bandaid.diff
Type: text/x-c
Size: 750 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20090920/08d485f9/attachment.bin>