suzanne.soy/racket
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

JIT: fix allocation of letrec-bound closure over unboxed flonums

Browse Source 
The closure could be allocated as uninitialized memory with the
expectation that it would be filled right away, but boxing values
to put in the closure could expose the uninitialized memory to
the GC. Fix the problem by boxing before allocating closures.
This commit is contained in:
Matthew Flatt 2014-10-01 13:04:50 -06:00
parent b336194cf4
commit b946d4639e
1 changed files with 10 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
racket/src/racket/src/jit.c
View File

@ -2846,6 +2846,13 @@ int scheme_generate(Scheme_Object *obj, mz_jit_state *jitter, int is_tail, int w
mz_rs_sync();
/* Box any unboxed values that will go into a closure */
for (i = 0; i < l->count; i++) {
if (generate_closure_prep((Scheme_Closure_Data *)l->procs[i], jitter))
prepped = 1;
CHECK_LIMIT();
}
/* Create unfinished closures */
for (i = 0; i < l->count; i++) {
((Scheme_Closure_Data *)l->procs[i])->context = (Scheme_Object *)l;
@ -2853,12 +2860,9 @@ int scheme_generate(Scheme_Object *obj, mz_jit_state *jitter, int is_tail, int w
CHECK_LIMIT();
jit_stxi_p(WORDS_TO_BYTES(i), JIT_RUNSTACK, JIT_R0);
}
for (i = 0; i < l->count; i++) {
if (generate_closure_prep((Scheme_Closure_Data *)l->procs[i], jitter))
prepped = 1;
CHECK_LIMIT();
}
/* We assume no allocation between last generated closure and
filling all closures, since the last one may be allocated as
"dirty". */
/* Close them: */
for (i = l->count; i--; ) {