diff options
author | David Malcolm <dmalcolm@redhat.com> | 2019-02-07 23:00:18 +0000 |
---|---|---|
committer | David Malcolm <dmalcolm@gcc.gnu.org> | 2019-02-07 23:00:18 +0000 |
commit | 25b67546a1a5fd49ddcb0a923a8d23d77a91e05f (patch) | |
tree | 9669c8a1154a56f24f9d2079c0759f4b6424e0ef /gcc/tree-vectorizer.h | |
parent | 8544ed6eea68a80999504c8a4b21b77d29cd86e2 (diff) | |
download | gcc-25b67546a1a5fd49ddcb0a923a8d23d77a91e05f.zip gcc-25b67546a1a5fd49ddcb0a923a8d23d77a91e05f.tar.gz gcc-25b67546a1a5fd49ddcb0a923a8d23d77a91e05f.tar.bz2 |
Fix more ICEs in -fsave-optimization-record (PR tree-optimization/89235)
PR tree-optimization/89235 reports an ICE inside -fsave-optimization-record
whilst reporting the inlining chain of of the location_t in the
vect_location global.
This is very similar to PR tree-optimization/86637, fixed in r266821.
The issue is that the inlining chains are read from the location_t's
ad-hoc data, referencing GC-managed tree blocks, but the former are
not GC roots; it's simply assumed that old locations referencing dead
blocks never get used again.
The fix is to reset the "vect_location" global in more places. Given
that is a somewhat subtle detail, the patch adds a sentinel class to
reset vect_location at the end of a scope. Doing it as a class
simplifies the task of ensuring that the global is reset on every
exit path from a function, and also gives a good place to signpost
the above subtlety (in the documentation for the class).
The patch also adds test cases for both of the PRs mentioned above.
gcc/testsuite/ChangeLog:
PR tree-optimization/86637
PR tree-optimization/89235
* gcc.c-torture/compile/pr86637-1.c: New test.
* gcc.c-torture/compile/pr86637-2.c: New test.
* gcc.c-torture/compile/pr86637-3.c: New test.
* gcc.c-torture/compile/pr89235.c: New test.
gcc/ChangeLog:
PR tree-optimization/86637
PR tree-optimization/89235
* tree-vect-loop.c (optimize_mask_stores): Add an
auto_purge_vect_location sentinel to ensure that vect_location is
purged on exit.
* tree-vectorizer.c
(auto_purge_vect_location::~auto_purge_vect_location): New dtor.
(try_vectorize_loop_1): Add an auto_purge_vect_location sentinel
to ensure that vect_location is purged on exit.
(pass_slp_vectorize::execute): Likewise, replacing the manual
reset.
* tree-vectorizer.h (class auto_purge_vect_location): New class.
From-SVN: r268659
Diffstat (limited to 'gcc/tree-vectorizer.h')
-rw-r--r-- | gcc/tree-vectorizer.h | 18 |
1 files changed, 18 insertions, 0 deletions
diff --git a/gcc/tree-vectorizer.h b/gcc/tree-vectorizer.h index d26b0f8..0e056f3 100644 --- a/gcc/tree-vectorizer.h +++ b/gcc/tree-vectorizer.h @@ -1420,6 +1420,24 @@ extern dump_user_location_t vect_location; #define DUMP_VECT_SCOPE(MSG) \ AUTO_DUMP_SCOPE (MSG, vect_location) +/* A sentinel class for ensuring that the "vect_location" global gets + reset at the end of a scope. + + The "vect_location" global is used during dumping and contains a + location_t, which could contain references to a tree block via the + ad-hoc data. This data is used for tracking inlining information, + but it's not a GC root; it's simply assumed that such locations never + get accessed if the blocks are optimized away. + + Hence we need to ensure that such locations are purged at the end + of any operations using them (e.g. via this class). */ + +class auto_purge_vect_location +{ + public: + ~auto_purge_vect_location (); +}; + /*-----------------------------------------------------------------*/ /* Function prototypes. */ /*-----------------------------------------------------------------*/ |