* Can't make a stack from a continuation
@ 2004-11-22 18:52 Neil Jerram
2004-11-25 19:43 ` Neil Jerram
0 siblings, 1 reply; 12+ messages in thread
From: Neil Jerram @ 2004-11-22 18:52 UTC (permalink / raw)
neil@laruns:~$ guile -q
guile> (version)
"1.6.4"
guile> (call-with-current-continuation make-stack)
Segmentation fault
This has been reported before, but it's still there. I think it's
something wrong with these lines from stacks.c, but I haven't
investigated further yet.
else if (SCM_CONTINUATIONP (obj))
{
offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (obj) + sizeof
(scm_t_contregs))
- SCM_BASE (obj));
#if SCM_STACK_GROWS_UP
offset += SCM_CONTINUATION_LENGTH (obj);
#endif
dframe = RELOC_FRAME (SCM_DFRAME (obj), offset);
}
Regards,
Neil
_______________________________________________
Bug-guile mailing list
Bug-guile@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-guile
^ permalink raw reply [flat|nested] 12+ messages in thread* Re: Can't make a stack from a continuation 2004-11-22 18:52 Can't make a stack from a continuation Neil Jerram @ 2004-11-25 19:43 ` Neil Jerram 2004-12-17 1:42 ` Neil Jerram 0 siblings, 1 reply; 12+ messages in thread From: Neil Jerram @ 2004-11-25 19:43 UTC (permalink / raw) Cc: bug-guile [-- Attachment #1: Type: text/plain, Size: 294 bytes --] Neil Jerram wrote: > neil@laruns:~$ guile -q > guile> (version) > "1.6.4" > guile> (call-with-current-continuation make-stack) > Segmentation fault I believe I have the fix for this (diffs attached for 1.6.x). Would anyone who feels half-confident in this area please review? Thanks, Neil [-- Attachment #2: stacks.c.diff --] [-- Type: text/x-patch, Size: 4239 bytes --] Index: libguile/stacks.c =================================================================== RCS file: /cvsroot/guile/guile/guile-core/libguile/stacks.c,v retrieving revision 1.64.2.4 diff -u -u -r1.64.2.4 stacks.c --- libguile/stacks.c 15 Mar 2002 10:33:37 -0000 1.64.2.4 +++ libguile/stacks.c 25 Nov 2004 19:43:20 -0000 @@ -162,10 +162,11 @@ if (SCM_EVALFRAMEP (*dframe)) { scm_t_debug_info * info = RELOC_INFO (dframe->info, offset); - n += (info - dframe->vect) / 2 + 1; + scm_t_debug_info * vect = RELOC_INFO (dframe->vect, offset); + n += (info - vect) / 2 + 1; /* Data in the apply part of an eval info frame comes from previous stack frame if the scm_t_debug_info vector is overflowed. */ - if ((((info - dframe->vect) & 1) == 0) + if ((((info - vect) & 1) == 0) && SCM_OVERFLOWP (*dframe) && !SCM_UNBNDP (info[1].a.proc)) ++n; @@ -174,7 +175,7 @@ ++n; } if (dframe && SCM_VOIDFRAMEP (*dframe)) - *id = dframe->vect[0].id; + *id = RELOC_INFO (dframe->vect, offset) -> id; else if (dframe) *maxp = 1; return n; @@ -189,7 +190,8 @@ if (SCM_EVALFRAMEP (*dframe)) { scm_t_debug_info * info = RELOC_INFO (dframe->info, offset); - if ((info - dframe->vect) & 1) + scm_t_debug_info * vect = RELOC_INFO (dframe->vect, offset); + if ((info - vect) & 1) { /* Debug.vect ends with apply info. */ --info; @@ -206,9 +208,10 @@ } else { + scm_t_debug_info * vect = RELOC_INFO (dframe->vect, offset); flags |= SCM_FRAMEF_PROC; - iframe->proc = dframe->vect[0].a.proc; - iframe->args = dframe->vect[0].a.args; + iframe->proc = vect[0].a.proc; + iframe->args = vect[0].a.args; } iframe->flags = flags; } @@ -254,6 +257,7 @@ { scm_t_info_frame *iframe = iframes; scm_t_debug_info *info; + scm_t_debug_info *vect; static SCM applybody = SCM_UNDEFINED; /* The value of applybody has to be setup after r4rs.scm has executed. */ @@ -275,7 +279,8 @@ --iframe; } info = RELOC_INFO (dframe->info, offset); - if ((info - dframe->vect) & 1) + vect = RELOC_INFO (dframe->vect, offset); + if ((info - vect) & 1) --info; /* Data in the apply part of an eval info frame comes from previous stack frame if the scm_t_debug_info vector is @@ -292,7 +297,7 @@ iframe->flags |= SCM_FRAMEF_OVERFLOW; info -= 2; NEXT_FRAME (iframe, n, quit); - while (info >= dframe->vect) + while (info >= vect) { if (!SCM_UNBNDP (info[1].a.proc)) { @@ -462,8 +467,7 @@ } else if (SCM_CONTINUATIONP (obj)) { - offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (obj) + sizeof (scm_t_contregs)) - - SCM_BASE (obj)); + offset = (SCM_CONTREGS (obj) -> stack) - SCM_BASE (obj); #ifndef STACK_GROWS_UP offset += SCM_CONTINUATION_LENGTH (obj); #endif @@ -490,7 +494,7 @@ SCM_STACK (stack) -> frames = iframe; /* Translate the current chain of stack frames into debugging information. */ - n = read_frames (RELOC_FRAME (dframe, offset), offset, n, iframe); + n = read_frames (dframe, offset, n, iframe); SCM_STACK (stack) -> length = n; /* Narrow the stack according to the arguments given to scm_make_stack. */ @@ -546,8 +550,7 @@ } else if (SCM_CONTINUATIONP (stack)) { - offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (stack) + sizeof (scm_t_contregs)) - - SCM_BASE (stack)); + offset = (SCM_CONTREGS (stack) -> stack) - SCM_BASE (stack); #ifndef STACK_GROWS_UP offset += SCM_CONTINUATION_LENGTH (stack); #endif @@ -565,7 +568,7 @@ while (dframe && !SCM_VOIDFRAMEP (*dframe)) dframe = RELOC_FRAME (dframe->prev, offset); if (dframe && SCM_VOIDFRAMEP (*dframe)) - return dframe->vect[0].id; + return RELOC_INFO (dframe->vect, offset) -> id; return SCM_BOOL_F; } #undef FUNC_NAME @@ -625,8 +628,7 @@ } else if (SCM_CONTINUATIONP (obj)) { - offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (obj) + sizeof (scm_t_contregs)) - - SCM_BASE (obj)); + offset = (SCM_CONTREGS (obj) -> stack) - SCM_BASE (obj); #ifndef STACK_GROWS_UP offset += SCM_CONTINUATION_LENGTH (obj); #endif [-- Attachment #3: eval.test.diff --] [-- Type: text/x-patch, Size: 899 bytes --] Index: test-suite/tests/eval.test =================================================================== RCS file: /cvsroot/guile/guile/guile-core/test-suite/tests/eval.test,v retrieving revision 1.6.2.1 diff -u -u -r1.6.2.1 eval.test --- test-suite/tests/eval.test 19 Jul 2001 20:49:34 -0000 1.6.2.1 +++ test-suite/tests/eval.test 25 Nov 2004 19:43:20 -0000 @@ -177,4 +177,26 @@ (map + '(1 2) '(3))) ))) +;;; +;;; continuations +;;; + +(with-test-prefix "continuation" + + (with-test-prefix "stacks/debugging" + + (debug-enable 'debug) + + (pass-if "make-stack" + (stack? (call-with-current-continuation make-stack))) + + (pass-if "stack-id" + (let ((id (call-with-current-continuation stack-id))) + (or (boolean? id) (symbol? id)))) + + (pass-if "last-stack-frame" + (pair? (call-with-current-continuation last-stack-frame))) + + )) + ;;; eval.test ends here [-- Attachment #4: Type: text/plain, Size: 137 bytes --] _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-11-25 19:43 ` Neil Jerram @ 2004-12-17 1:42 ` Neil Jerram 2004-12-23 14:24 ` Marius Vollmer 0 siblings, 1 reply; 12+ messages in thread From: Neil Jerram @ 2004-12-17 1:42 UTC (permalink / raw) Cc: bug-guile Neil Jerram wrote: > Neil Jerram wrote: > >> neil@laruns:~$ guile -q >> guile> (version) >> "1.6.4" >> guile> (call-with-current-continuation make-stack) >> Segmentation fault > > > I believe I have the fix for this (diffs attached for 1.6.x). Would > anyone who feels half-confident in this area please review? A further aspect of this that has been bothering me, is why it doesn't show up in all the occurrences of (make-stack *cont*) that my debugging code does when called from one of the eval trap handlers. The answer (I believe) is that the eval trap handlers don't use real continuations; they use "debug objects" instead; this is controlled by the 'cheaptraps debug option, which defaults to "yes". And one more problem ... the same problem exists in CVS head, but a similar patch doesn't fix it ... still investigating. Neil _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-17 1:42 ` Neil Jerram @ 2004-12-23 14:24 ` Marius Vollmer 2004-12-23 15:36 ` Marius Vollmer 0 siblings, 1 reply; 12+ messages in thread From: Marius Vollmer @ 2004-12-23 14:24 UTC (permalink / raw) Cc: bug-guile, guile-devel Neil Jerram <neil@ossau.uklinux.net> writes: > And one more problem ... the same problem exists in CVS head, but a > similar patch doesn't fix it ... still investigating. I'm looking into this as well, now. The fix you did for 1.6 looks good, although I don't understand it completely yet. I will try to come up with my own patch to CVS head. One larger change I have in mind is to move the offset calculation into the continuation code itself, which is where it belongs. Each continuation will get a new "offset" field and the stack code will just use it. _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-23 14:24 ` Marius Vollmer @ 2004-12-23 15:36 ` Marius Vollmer 2004-12-24 23:05 ` Neil Jerram 0 siblings, 1 reply; 12+ messages in thread From: Marius Vollmer @ 2004-12-23 15:36 UTC (permalink / raw) Cc: bug-guile, guile-devel Marius Vollmer <marius.vollmer@uni-dortmund.de> writes: > Neil Jerram <neil@ossau.uklinux.net> writes: > >> And one more problem ... the same problem exists in CVS head, but a >> similar patch doesn't fix it ... still investigating. > > I'm looking into this as well, now. The fix you did for 1.6 looks > good, although I don't understand it completely yet. I will try to > come up with my own patch to CVS head. Here it is. I have already applied it to CVS. I think a similar scheme can be used for 1.6, too: i.e., move the offset calculation into scm_make_continuation. Neil, do you want to fix 1.6 yourself? Index: continuations.c =================================================================== RCS file: /cvsroot/guile/guile/guile-core/libguile/continuations.c,v retrieving revision 1.54 diff -u -r1.54 continuations.c --- continuations.c 22 Oct 2004 15:13:12 -0000 1.54 +++ continuations.c 23 Dec 2004 15:26:20 -0000 @@ -134,6 +134,7 @@ #if ! SCM_STACK_GROWS_UP src -= stack_size; #endif + continuation->offset = continuation->stack - src; memcpy (continuation->stack, src, sizeof (SCM_STACKITEM) * stack_size); #ifdef __ia64__ Index: continuations.h =================================================================== RCS file: /cvsroot/guile/guile/guile-core/libguile/continuations.h,v retrieving revision 1.31 diff -u -r1.31 continuations.h --- continuations.h 25 Jun 2003 18:12:35 -0000 1.31 +++ continuations.h 23 Dec 2004 15:26:20 -0000 @@ -55,8 +55,19 @@ size_t num_stack_items; /* size of the saved stack. */ unsigned long seq; /* dynamic root identifier. */ - /* the most recently created debug frame on the live stack, before - it was saved. */ + /* The offset from the live stack location and this copy. This is + used to adjust pointers from within the copied stack to the stack + itself. + + Thus, when you read a pointer from the copied stack that points + into the live stack, you need to add OFFSET so that it points + into the copy. + */ + scm_t_ptrdiff offset; + + /* The most recently created debug frame on the live stack, before + it was saved. This need to be adjusted with OFFSET, above. + */ struct scm_t_debug_frame *dframe; SCM_STACKITEM stack[1]; /* copied stack of size num_stack_items. */ Index: stacks.c =================================================================== RCS file: /cvsroot/guile/guile/guile-core/libguile/stacks.c,v retrieving revision 1.80 diff -u -r1.80 stacks.c --- stacks.c 22 Sep 2004 17:41:37 -0000 1.80 +++ stacks.c 23 Dec 2004 15:26:20 -0000 @@ -124,7 +124,8 @@ * is read from a continuation. */ static scm_t_bits -stack_depth (scm_t_debug_frame *dframe, long offset, SCM *id, int *maxp) +stack_depth (scm_t_debug_frame *dframe, scm_t_ptrdiff offset, + SCM *id, int *maxp) { long n; long max_depth = SCM_BACKTRACE_MAXDEPTH; @@ -134,11 +135,12 @@ { if (SCM_EVALFRAMEP (*dframe)) { - scm_t_debug_info * info = RELOC_INFO (dframe->info, offset); - n += (info - dframe->vect) / 2 + 1; + scm_t_debug_info *info = RELOC_INFO (dframe->info, offset); + scm_t_debug_info *vect = RELOC_INFO (dframe->vect, offset); + n += (info - vect) / 2 + 1; /* Data in the apply part of an eval info frame comes from previous stack frame if the scm_t_debug_info vector is overflowed. */ - if ((((info - dframe->vect) & 1) == 0) + if ((((info - vect) & 1) == 0) && SCM_OVERFLOWP (*dframe) && !SCM_UNBNDP (info[1].a.proc)) ++n; @@ -147,7 +149,7 @@ ++n; } if (dframe && SCM_VOIDFRAMEP (*dframe)) - *id = dframe->vect[0].id; + *id = RELOC_INFO(dframe->vect, offset)[0].id; else if (dframe) *maxp = 1; return n; @@ -156,13 +158,15 @@ /* Read debug info from DFRAME into IFRAME. */ static void -read_frame (scm_t_debug_frame *dframe, long offset, scm_t_info_frame *iframe) +read_frame (scm_t_debug_frame *dframe, scm_t_ptrdiff offset, + scm_t_info_frame *iframe) { scm_t_bits flags = SCM_UNPACK (SCM_INUM0); /* UGh. */ if (SCM_EVALFRAMEP (*dframe)) { - scm_t_debug_info * info = RELOC_INFO (dframe->info, offset); - if ((info - dframe->vect) & 1) + scm_t_debug_info *info = RELOC_INFO (dframe->info, offset); + scm_t_debug_info *vect = RELOC_INFO (dframe->vect, offset); + if ((info - vect) & 1) { /* Debug.vect ends with apply info. */ --info; @@ -179,9 +183,10 @@ } else { + scm_t_debug_info *vect = RELOC_INFO (dframe->vect, offset); flags |= SCM_FRAMEF_PROC; - iframe->proc = dframe->vect[0].a.proc; - iframe->args = dframe->vect[0].a.args; + iframe->proc = vect[0].a.proc; + iframe->args = vect[0].a.args; } iframe->flags = flags; } @@ -223,10 +228,11 @@ */ static scm_t_bits -read_frames (scm_t_debug_frame *dframe, long offset, long n, scm_t_info_frame *iframes) +read_frames (scm_t_debug_frame *dframe, scm_t_ptrdiff offset, + long n, scm_t_info_frame *iframes) { scm_t_info_frame *iframe = iframes; - scm_t_debug_info *info; + scm_t_debug_info *info, *vect; static SCM applybody = SCM_UNDEFINED; /* The value of applybody has to be setup after r4rs.scm has executed. */ @@ -248,7 +254,8 @@ --iframe; } info = RELOC_INFO (dframe->info, offset); - if ((info - dframe->vect) & 1) + vect = RELOC_INFO (dframe->vect, offset); + if ((info - vect) & 1) --info; /* Data in the apply part of an eval info frame comes from previous stack frame if the scm_t_debug_info vector is @@ -265,7 +272,7 @@ iframe->flags |= SCM_FRAMEF_OVERFLOW; info -= 2; NEXT_FRAME (iframe, n, quit); - while (info >= dframe->vect) + while (info >= vect) { if (!SCM_UNBNDP (info[1].a.proc)) { @@ -435,12 +442,9 @@ } else if (SCM_CONTINUATIONP (obj)) { - offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (obj) + sizeof (scm_t_contregs)) - - SCM_BASE (obj)); -#if SCM_STACK_GROWS_UP - offset += SCM_CONTINUATION_LENGTH (obj); -#endif - dframe = RELOC_FRAME (SCM_DFRAME (obj), offset); + scm_t_contregs *cont = SCM_CONTREGS (obj); + offset = cont->offset; + dframe = RELOC_FRAME (cont->dframe, offset); } else { @@ -463,7 +467,7 @@ SCM_STACK (stack) -> frames = iframe; /* Translate the current chain of stack frames into debugging information. */ - n = read_frames (RELOC_FRAME (dframe, offset), offset, n, iframe); + n = read_frames (dframe, offset, n, iframe); SCM_STACK (stack) -> length = n; /* Narrow the stack according to the arguments given to scm_make_stack. */ @@ -519,12 +523,9 @@ } else if (SCM_CONTINUATIONP (stack)) { - offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (stack) + sizeof (scm_t_contregs)) - - SCM_BASE (stack)); -#if SCM_STACK_GROWS_UP - offset += SCM_CONTINUATION_LENGTH (stack); -#endif - dframe = RELOC_FRAME (SCM_DFRAME (stack), offset); + scm_t_contregs *cont = SCM_CONTREGS (stack); + offset = cont->offset; + dframe = RELOC_FRAME (cont->dframe, offset); } else if (SCM_STACKP (stack)) { @@ -538,7 +539,7 @@ while (dframe && !SCM_VOIDFRAMEP (*dframe)) dframe = RELOC_FRAME (dframe->prev, offset); if (dframe && SCM_VOIDFRAMEP (*dframe)) - return dframe->vect[0].id; + return RELOC_INFO (dframe->vect, offset)[0].id; return SCM_BOOL_F; } #undef FUNC_NAME @@ -595,12 +596,9 @@ } else if (SCM_CONTINUATIONP (obj)) { - offset = ((SCM_STACKITEM *) ((char *) SCM_CONTREGS (obj) + sizeof (scm_t_contregs)) - - SCM_BASE (obj)); -#if SCM_STACK_GROWS_UP - offset += SCM_CONTINUATION_LENGTH (obj); -#endif - dframe = RELOC_FRAME (SCM_DFRAME (obj), offset); + scm_t_contregs *cont = SCM_CONTREGS (obj); + offset = cont->offset; + dframe = RELOC_FRAME (cont->dframe, offset); } else { _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-23 15:36 ` Marius Vollmer @ 2004-12-24 23:05 ` Neil Jerram 2004-12-24 23:11 ` Neil Jerram 2004-12-24 23:21 ` Marius Vollmer 0 siblings, 2 replies; 12+ messages in thread From: Neil Jerram @ 2004-12-24 23:05 UTC (permalink / raw) Cc: bug-guile, guile-devel Marius Vollmer wrote: > Marius Vollmer <marius.vollmer@uni-dortmund.de> writes: > > >>Neil Jerram <neil@ossau.uklinux.net> writes: >> >> >>>And one more problem ... the same problem exists in CVS head, but a >>>similar patch doesn't fix it ... still investigating. >> >>I'm looking into this as well, now. The fix you did for 1.6 looks >>good, although I don't understand it completely yet. I will try to >>come up with my own patch to CVS head. > > > Here it is. I have already applied it to CVS. I think a similar > scheme can be used for 1.6, too: i.e., move the offset calculation > into scm_make_continuation. > > Neil, do you want to fix 1.6 yourself? > > [patch] This change (calculating and storing the offset in scm_make_continuation) makes sense, but I'd like also to understand where the bug was in my version (using offset = (SCM_CONTREGS (stack) -> stack) - SCM_BASE (stack);) and why my code worked in 1.6 but not in head. Would you mind explaining? Thanks, Neil PS. Merry Christmas! _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-24 23:05 ` Neil Jerram @ 2004-12-24 23:11 ` Neil Jerram 2004-12-24 23:22 ` Marius Vollmer 2004-12-24 23:21 ` Marius Vollmer 1 sibling, 1 reply; 12+ messages in thread From: Neil Jerram @ 2004-12-24 23:11 UTC (permalink / raw) Cc: bug-guile, guile-devel Neil Jerram wrote: > This change (calculating and storing the offset in > scm_make_continuation) makes sense [...] One more thing - are you happy with the proposed new tests for 1.6 and head? I think one change is needed, namely to save and restore (debug-options) so that the effect of (debug-enable 'debug) can't affect other tests. Neil _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-24 23:11 ` Neil Jerram @ 2004-12-24 23:22 ` Marius Vollmer 2004-12-25 10:51 ` Neil Jerram 0 siblings, 1 reply; 12+ messages in thread From: Marius Vollmer @ 2004-12-24 23:22 UTC (permalink / raw) Cc: bug-guile, guile-devel Neil Jerram <neil@ossau.uklinux.net> writes: > One more thing - are you happy with the proposed new tests for 1.6 and > head? Yes. > I think one change is needed, namely to save and restore > (debug-options) so that the effect of (debug-enable 'debug) can't > affect other tests. Sounds good to me, too. Could you add this? -- GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3 331E FAF8 226A D5D4 E405 _______________________________________________ Guile-devel mailing list Guile-devel@gnu.org http://lists.gnu.org/mailman/listinfo/guile-devel ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-24 23:22 ` Marius Vollmer @ 2004-12-25 10:51 ` Neil Jerram 2004-12-27 23:23 ` Neil Jerram 0 siblings, 1 reply; 12+ messages in thread From: Neil Jerram @ 2004-12-25 10:51 UTC (permalink / raw) Cc: bug-guile, guile-devel Marius Vollmer wrote: > Neil Jerram <neil@ossau.uklinux.net> writes: > > >>One more thing - are you happy with the proposed new tests for 1.6 and >>head? > > > Yes. > > >>I think one change is needed, namely to save and restore >>(debug-options) so that the effect of (debug-enable 'debug) can't >>affect other tests. > > > Sounds good to me, too. Could you add this? Yes, will do. Probably not today, though :-) Neil _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-25 10:51 ` Neil Jerram @ 2004-12-27 23:23 ` Neil Jerram 0 siblings, 0 replies; 12+ messages in thread From: Neil Jerram @ 2004-12-27 23:23 UTC (permalink / raw) Cc: bug-guile, guile-devel Neil Jerram wrote: > > Yes, will do. Probably not today, though :-) This is now complete. Two notes: - In HEAD, I moved the new tests from eval.test to continuations.test, as the latter seems more appropriate. (I only used eval.test in 1.6 in order to avoid having to create a new file.) - In 1.6, scm_t_ptrdiff is not available, so I used long instead. Regards, Neil _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-24 23:05 ` Neil Jerram 2004-12-24 23:11 ` Neil Jerram @ 2004-12-24 23:21 ` Marius Vollmer 2004-12-25 10:50 ` Neil Jerram 1 sibling, 1 reply; 12+ messages in thread From: Marius Vollmer @ 2004-12-24 23:21 UTC (permalink / raw) Cc: bug-guile, guile-devel Neil Jerram <neil@ossau.uklinux.net> writes: > This change (calculating and storing the offset in > scm_make_continuation) makes sense, but I'd like also to understand > where the bug was in my version (using offset = (SCM_CONTREGS > (stack) -> stack) - SCM_BASE (stack);) and why my code worked in 1.6 > but not in head. Would you mind explaining? Hmm, I didn't really try to debug your patch. I followed its idea of also relocating dframe->vect and made the changes that seemed right. Using offset = SCM_CONTREGS (stack)->stack - SCM_BASE(stack) looks right. The bug is probably elsewhere... > PS. Merry Christmas! Thanks and Merry Christmas from me to everyone as well! (I got a high-tech electric iron in transparent iMac design from my mom! :-) -- GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3 331E FAF8 226A D5D4 E405 _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
* Re: Can't make a stack from a continuation 2004-12-24 23:21 ` Marius Vollmer @ 2004-12-25 10:50 ` Neil Jerram 0 siblings, 0 replies; 12+ messages in thread From: Neil Jerram @ 2004-12-25 10:50 UTC (permalink / raw) Cc: bug-guile, guile-devel Marius Vollmer wrote: > > Hmm, I didn't really try to debug your patch. I followed its idea of > also relocating dframe->vect and made the changes that seemed right. > > Using offset = SCM_CONTREGS (stack)->stack - SCM_BASE(stack) looks > right. The bug is probably elsewhere... Fair enough, but when I compared your and my patches, they were identical except for - the offset calculation and storage - two occurrences of x->id instead of x[0].id. And I thought that mine didn't work ... Anyway, as long as yours does work now, it's probably not worth worrying about this further; it may be that I had an inconsistent build, or LD_LIBRARY not set properly, or something. Neil _______________________________________________ Bug-guile mailing list Bug-guile@gnu.org http://lists.gnu.org/mailman/listinfo/bug-guile ^ permalink raw reply [flat|nested] 12+ messages in thread
end of thread, other threads:[~2004-12-27 23:23 UTC | newest] Thread overview: 12+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2004-11-22 18:52 Can't make a stack from a continuation Neil Jerram 2004-11-25 19:43 ` Neil Jerram 2004-12-17 1:42 ` Neil Jerram 2004-12-23 14:24 ` Marius Vollmer 2004-12-23 15:36 ` Marius Vollmer 2004-12-24 23:05 ` Neil Jerram 2004-12-24 23:11 ` Neil Jerram 2004-12-24 23:22 ` Marius Vollmer 2004-12-25 10:51 ` Neil Jerram 2004-12-27 23:23 ` Neil Jerram 2004-12-24 23:21 ` Marius Vollmer 2004-12-25 10:50 ` Neil Jerram
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).