* libguile/gc.c (scm_oom_fn, scm_init_gc): Install an out-of-memory
handler that raises an unwind-only out-of-memory exception.
(scm_gc_warn_proc, scm_init_gc): Install a warning proc that tries to
print to the current warning port, if the current warning port is a
file port.
(scm_gc_after_nonlocal_exit): New interface. Should be called after a
nonlocal return to potentially collect memory; otherwise allocations
could try to expand again when they should collect.
* libguile/continuations.c (scm_i_make_continuation):
* libguile/eval.c (eval):
* libguile/throw.c (catch):
* libguile/vm.c (scm_call_n): Call scm_gc_after_nonlocal_exit after
nonlocal returns.
* libguile/throw.c (abort_to_prompt, throw_without_pre_unwind): Rework
to avoid allocating memory.
(scm_report_out_of_memory): New interface.
(scm_init_throw): Pre-allocate the arguments for stack-overflow and
out-of-memory errors.
* module/ice-9/boot-9.scm: Add an out-of-memory exception printer.
* module/system/repl/error-handling.scm (call-with-error-handling): Add
out-of-memory to the report-keys set.
* libguile/gc-malloc.c (scm_realloc): Call scm_report_out_of_memory if
realloc fails.
* libguile/error.h:
* libguile/error.c:
* libguile/deprecated.h:
* libguile/deprecated.c (scm_memory_error): Deprecate.
* test-suite/standalone/Makefile.am:
* test-suite/standalone/test-out-of-memory: New test case.
* libguile/throw.c (throw_without_pre_unwind): Newline after the
unwind-only warning.
* test-suite/standalone/Makefile.am:
* test-suite/standalone/test-stack-overflow: New test to handle
mmap/malloc failure.
* module/ice-9/boot-9.scm (catch): Signal an early error if the handler
or pre-unwind handler types aren't right. This is more important than
it was, given that we dispatch on type now when finding matching catch
clauses.
* libguile/vm.c (vm_expand_stack): Use the standard
scm_report_stack_overflow to signal stack overflow. This will avoid
running pre-unwind handlers.
* libguile/throw.h: Move scm_report_stack_overflow here.
* libguile/throw.c (catch): Define a version of catch in C.
(throw_without_pre_unwind): New helper. Besides serving as the
pre-boot "throw" binding, it allows stack overflow to throw without
running pre-unwind handlers.
(scm_catch, scm_catch_with_pre_unwind_handler)
(scm_with_throw_handler): Use the new catch in C.
(scm_report_stack_overflow): Moved from stackchk.c; throws an
unwind-only exception.
* libguile/stackchk.h:
* libguile/stackchk.c: Remove the scm_report_stack_overflow bits.
* libguile/throw.c (scm_init_throw): Define %exception-handler here.
* module/ice-9/boot-9.scm (%eh): Use the incoming %exception-handler,
and then delete it. This way we should be able to do unwind-only
exceptions from C.
* libguile/control.h:
* libguile/control.c (scm_i_prompt_pop_abort_args_x): Change to take VP
as an arg, not VM.
* libguile/eval.c (eval):
* libguile/throw.c (pre_init_catch): Adapt.
* libguile/dynstack.h:
* libguile/dynstack.c (PROMPT_FP, PROMPT_SP):
(scm_dynstack_push_prompt, scm_dynstack_find_prompt): Prompts on the
dynstack are recorded as offsets from the base stack address in this
thread.
* libguile/control.c (scm_c_abort):
* libguile/eval.c (eval):
* libguile/stacks.c (find_prompt, narrow_stack):
* libguile/throw.c (pre_init_catch):
* libguile/vm-engine.c (prompt): Adapt.
Fixes <http://bugs.gnu.org/15798>.
Reported by Matt Sicker <boards@gmail.com>.
* libguile/__scm.h (SCM_NORETURN): Use ((__noreturn__)) instead of
((noreturn)).
* libguile/throw.h (scm_ithrow): Rename formal parameter from 'noreturn'
to 'no_return'.
* libguile/throw.c (scm_ithrow): Rename formal parameter from 'noreturn'
to 'no_return'.
* libguile/dynstack.h (scm_t_dynstack_prompt_flags): New flag,
SCM_F_DYNSTACK_PROMPT_PUSH_NARGS, set if the continuation expects the
number of args to be pushed on the top of the stack.
* libguile/control.c (scm_c_abort): Only push nargs if requested.
* libguile/eval.c (eval):
* libguile/throw.c (pre_init_catch):
* libguile/vm-i-system.c (prompt): Set
SCM_F_DYNSTACK_PROMPT_PUSH_NARGS.
* libguile/memoize.h:
* libguile/memoize.c (MAKMEMO_CALL_WITH_PROMPT, memoize, unmemoize):
Remove the @prompt memoizer in favor of recognizing call-with-prompt
primcalls. Rename SCM_M_PROMPT to SCM_M_CALL_WITH_PROMPT, and pass a
thunk instead of an expression so that it has normal applicative
order.
* libguile/expand.c (PRIMITIVE_REF, PRIMCALL, expand): Produce primcalls
from forms whose car is a primitive.
(expand_atat): Recognize (@@ primitive FOO) as being a primitive-ref.
* module/ice-9/boot-9.scm (call-with-prompt): Instead of dispatching to
the wonky @prompt memoizer, residualize a primcall to
call-with-prompt. The memoizer will DTRT to allow call-with-prompt to
be interpreted correctly without needing an additional binding.
* module/ice-9/eval.scm (primitive-eval): Change the 'prompt clause to a
call to call-with-prompt.
* module/language/tree-il/primitives.scm: No more need to recognize
@prompt.
* libguile/eval.c (eval): Adapt to SCM_M_PROMPT renaming to
SCM_M_CALL_WITH_PROMPT, and apply the thunk.
* libguile/throw.c (pre_init_throw): Adapt to scm_abort_to_prompt_star
rename.
* libguile/control.h: Remove scm_t_prompt_registers and
scm_c_make_prompt_registers.
(scm_c_abort): Take a pointer to a jmpbuf instead of a cookie. It
will serve the same purpose.
* libguile/control.c (reify_partial_continuation, scm_at_abort): Adapt
to new prompt representation.
* libguile/dynstack.h:
* libguile/dynstack.c (scm_dynstack_push_prompt): Prompts now have 5
words instead of 2, as they now push the fp, sp, ip, and jmpbuf on the
stack separately. This avoids allocation.
(scm_dynstack_find_prompt): Likewise, add return values for fp, sp,
etc.
(scm_dynstack_wind_prompt): Replaces scm_dynstack_relocate_prompt.
* libguile/eval.c (eval):
* libguile/stacks.c (find_prompt):
* libguile/throw.c (pre_init_catch): Adapt to the new prompt mechanism.
* libguile/vm-engine.c (vm_engine): Setjmp an on-stack jmpbuf every time
the VM enters. We can then re-use that jmpbuf for all prompts in that
invocation.
* libguile/vm-i-system.c (partial_cont_call): Adapt to change in prompt
representation. We don't need to wind here any more, since we pass in
the prompt's jmpbuf.
(prompt): Adapt to scm_dynstack_push_prompt change.
(abort): Adapt to vm_abort change.
* libguile/vm.h (struct scm_vm): No more cookie.
* libguile/vm.c (vm_abort): Adapt to scm_c_abort change.
(vm_reinstate_partial_continuation): Rewind the dynamic stack here,
now that we do have a valid jmpbuf.
(make_vm): No need to initialize a cookie.
* libguile/dynstack.h:
* libguile/dynstack.c: New files, implementing the dynamic stack as a
true stack instead of a linked list. This lowers the cost of
dynwinds: frames, winders, prompts, with-fluids, and dynamic-wind.
For the most part, we allocate these items directly on the stack.
* libguile/dynwinds.h:
* libguile/dynwinds.c: Adapt all manipulators of the wind stack to use
interfaces from dynstack.c. Remove heap-allocated winder and frame
object types.
(scm_dowinds, scm_i_dowinds): Remove these. The first was exported,
but it was not a public interface.
* libguile/continuations.c:
* libguile/continuations.h (scm_t_contregs): Continuation objects
reference scm_t_dynstack* values now. Adapt to the new interfaces.
* libguile/control.c:
* libguile/control.h: There is no longer a scm_tc7_prompt kind of object
that can be allocated on the heap. Instead, the prompt flags, key,
and registers are pushed on the dynwind stack. (The registers are
still on the heap.) Also, since the vm_cont will reference the
dynwinds, make the partial continuation stub take just one extra arg,
instead of storing the intwinds separately in the object table.
* libguile/fluids.c:
* libguile/fluids.h: No more with-fluids objects; instead, the fluids go
on the dynstack. The values still have to be on the heap, though.
(scm_prepare_fluids, scm_swap_fluids): New internal functions,
replacing scm_i_make_with_fluids and scm_i_swap_with_fluids.
* libguile/print.c: Remove prompt and with-fluids printers.
* libguile/tags.h: Revert prompt and with-fluids tc7 values to what they
were before they were allocated.
* libguile/vm-i-system.c (partial_cont_call): Just pop the vmcont, the
intwinds will not be passed as a second arg. Rewind the dynamic stack
from within the VM, so that any rewinder sees valid prompt entries.
(call_cc, tail_call_cc): Adapt to pass the dynstack to
scm_i_vm_capture_stack.
(prompt, wind, unwind, wind_fluids, unwind_fluids): Adapt to the new
interfaces.
* libguile/vm.h (scm_i_capture_current_stack): Rename from
scm_i_vm_capture_continuation.
(scm_i_vm_capture_stack): Take a dynstack as an argument.
* libguile/vm.c (vm_reinstate_partial_continuation): Don't wind here, as
that could result in winders seeing invalid prompts.
* libguile/eval.c:
* libguile/root.c:
* libguile/stacks.c:
* libguile/threads.c:
* libguile/threads.h:
* libguile/throw.c: Adapt other users of dynwinds to use the dynstack.
* libguile/ports.c (scm_putc, scm_puts):
* libguile/ports.h (scm_putc_unlocked, scm_puts_unlocked): Separate into
_unlocked and locked variants. Change all callers to use the
_unlocked versions.
* libguile/throw.c (scm_exit_status): In one of my bogus patches, I
managed to make #<unspecified> reach this function, causing a segfault
on SCM_CAR. Refactor to be more robust.
* libguile/control.h:
* libguile/control.c (scm_i_prompt_pop_abort_args_x): Take a VM instead
of a prompt, given that it's the VM's registers that record the abort
arguments, not the prompt registers (which actually point right below
the abort values).
* libguile/eval.c (eval):
* libguile/throw.c (pre_init_catch): Pass the vm instead of a prompt.
* libguile/backtrace.c (scm_print_exception): Use scm_module_variable to
look up print-exception so that it works before boot-9 is loaded.
* libguile/throw.c (CACHE_VAR): Tweak to use scm_from_latin1_symbol.
* libguile/throw.c (handler_message, should_print_backtrace): Use
scm_print_exception. Add a helper function to determine when to print
a backtrace; don't do so on read or syntax errors.
* libguile/backtrace.c (display_error_body): Don't look up the
frame-source if Guile isn't initialized yet. Fixes display-error
before boot has finished.
* libguile/throw.c (handler_message): Likewise, don't backtrace before
boot has finished, because we can't load (system vm frame).
* libguile/throw.c (find_pre_init_catch): New internal helper.
(pre_init_throw): If we don't find a catch, print out a message and
abort.
(scm_init_throw): Declare pre-init-throw as taking at least one arg.
* libguile/bytevectors.c:
* libguile/eval.c:
* libguile/goops.c:
* libguile/i18n.c:
* libguile/load.c:
* libguile/memoize.c:
* libguile/modules.c:
* libguile/ports.c:
* libguile/print.c:
* libguile/procs.c:
* libguile/programs.c:
* libguile/read.c:
* libguile/script.c:
* libguile/srfi-14.c:
* libguile/stacks.c:
* libguile/strings.c:
* libguile/throw.c:
* libguile/vm.c: Use scm_from_latin1_symboln to make symbols from string
literals, because they aren't in the user's locale -- they are in
ASCII, and we can optimize this case.
* libguile/vm-i-loader.c: Also use scm_from_latin1_symboln when loading
narrow symbols.
* module/system/repl/repl.scm (display-syntax-error): New helper,
displays a syntax error.
(abort-on-error, run-repl): Use it.
* libguile/throw.c (handler_message): Re-code the same thing in C.
* libguile/backtrace.h (scm_the_last_stack_fluid_var)
* libguile/backtrace.c (scm_init_backtrace): No more
scm_the_last_stack_fluid_var. The replacement is to resolve
`the-last-stack' in (ice-9 stack-catch).
(scm_backtrace_with_highlights): Accordingly, instead of backtracing
the last stack, backtrace the current stack.
* libguile/throw.h:
* libguile/throw.c:
* libguile/deprecated.h:
* libguile/deprecated.c (scm_internal_stack_catch): Deprecate this
function.
* module/ice-9/save-stack.scm (the-last-stack): Move here from boot-9.
* module/ice-9/debug.scm:
* module/ice-9/debugger.scm: Use (ice-9 save-stack) for the-last-stack.
* module/ice-9/deprecated.scm (the-last-stack): Add deprecated shim.
* libguile/control.h (scm_sys_default_prompt_tag):
* libguile/control.c (scm_init_control): Remove the logic that defined
%default-prompt-tag.
(scm_c_abort): Remove check for default prompt tag, it wasn't useful.
* libguile/throw.c (sym_pre_init_catch_tag): Define as the pre-init
prompt tag.
(pre_init_catch, pre_init_throw): Use sym_pre_init_catch_tag.
* module/ice-9/boot-9.scm (default-prompt-tag): Define as a simple
value, not a fluid. Perhaps we can expose it as a fluid later.
* libguile/control.h (SCM_PROMPT_HANDLER): Remove, it was unused.
(SCM_PROMPT_DYNWINDS): Rename from SCM_PROMPT_DYNENV.
* libguile/control.c: (scm_c_make_prompt): Take another arg, the winds
that are to be in place for the prompt. Fix allocation to be 4 words
instead of 5 (the handler was never used).
* libguile/eval.c (eval):
* libguile/throw.c (pre_init_catch): Adapt to scm_c_make_prompt change.
* libguile/vm-i-system.c (partial-cont-call): Grovel the new elements of
the wind list in order to call setjmp() on the new prompts. Pass
cookie to vm_reinstate_partial_continuation.
(prompt): Adapt to scm_c_make_prompt change.
* libguile/vm.c (vm_reinstate_partial_continuation): Take a cookie arg,
used when winding captured prompts onto the stack. Winding a prompt
implies making a new prompt, actually -- with new registers, a new
jump buffer, new winds, etc.
* test-suite/tests/control.test ("rewinding prompts"): Add a test for
rewinding prompts.
* libguile/throw.c (tc16_jmpbuffer, tc16_pre_unwind_data): Remove these
smob types, and associated constructors and accessors (all internal).
(scm_catch, scm_catch_with_pre_unwind_handler):
(scm_with_throw_handler, scm_throw): Simply dispatch to scheme.
Lovely.
(tc16_catch_closure): Introduce a new applicable smob type, for use by
the C catch interface. All constructors and accessors are internal.
(scm_c_catch, scm_internal_catch, scm_c_with_throw_handler): Build
applicable smobs out of the C procedure arguments, so we can then
dispatch through scm_catch et al.
(scm_ithrow): Dispatch to scm_throw.
(pre_init_catch, pre_init_throw): Restricted catch/throw
implementation for use before boot-9 runs.
(scm_init_throw): Bind the pre-init catch and throw definitions.
* module/ice-9/boot-9.scm (prompt, abort): Move these definitions up in
the file.
(catch, throw, with-throw-handler): Implement in Scheme. Whee!
* libguile/deprecated.h:
* libguile/deprecated.c (scm_internal_lazy_catch, scm_lazy_catch):
Deprecate, and print out a nasty warning that people should change to
with-throw-handler.
* libguile/throw.h:
* libguile/throw.c (scm_c_with_throw_handler): Deprecate the use of the
lazy_catch_p argument, printing out a nasty warning if someone
actually passes 1 as that argument. The combination of the pre-unwind
and post-unwind handlers should be sufficient.
* test-suite/tests/exceptions.test: Remove lazy-catch tests, as they are
deprecated. Two of them fail:
* throw/catch: effect of lazy-catch unwinding on throw to another key
* throw/catch: repeat of previous test but with lazy-catch
Hopefully people are not depending on this behavior, and the warning is
sufficiently nasty for people to switch. We will see.
* test-suite/tests/eval.test ("promises"): Use with-throw-handler
instead of lazy-catch.
* doc/ref/api-debug.texi:
* doc/ref/api-control.texi: Update to remove references to lazy-catch,
folding in the useful bits to with-throw-handler.
* libguile/debug.h (scm_t_debug_frame): Remove this type, as it was
internal to the old evaluator.
(SCM_EVALFRAME, SCM_APPLYFRAME, SCM_VOIDFRAME, SCM_MACROEXPF)
(SCM_TAILREC, SCM_TRACED_FRAME, SCM_ARGS_READY, SCM_DOVERFLOW)
(SCM_MAX_FRAME_SIZE, SCM_FRAMETYPE)
(SCM_EVALFRAMEP, SCM_APPLYFRAMEP, SCM_VOIDFRAMEP, SCM_MACROEXPFP)
(SCM_TAILRECP, SCM_TRACED_FRAME_P, SCM_ARGS_READY_P, SCM_OVERFLOWP)
(SCM_SET_MACROEXP, SCM_SET_TAILREC, SCM_SET_TRACED_FRAME)
(SCM_SET_ARGSREADY, SCM_SET_OVERFLOW)
(SCM_CLEAR_MACROEXP, SCM_CLEAR_TRACED_FRAME, SCM_CLEAR_ARGSREADY):
Remove macro accessors to scm_t_debug_frame.
(SCM_DEBUGOBJP, SCM_DEBUGOBJ_FRAME, SCM_SET_DEBUGOBJ_FRAME):
(scm_debug_object_p, scm_make_debugobj): Remove debugobj accessors.
(scm_i_unmemoize_expr): Remove unused declaration.
* libguile/debug.c (scm_debug_options): No more max limit on frame
sizes.
(scm_start_stack): Just call out to scm_vm_call_with_new_stack.
(scm_debug_object_p, scm_make_debugobj, scm_init_debug): No more
debugobj smob type.
* libguile/deprecated.h:
* libguile/deprecated.c (scm_i_deprecated_last_debug_frame)
(scm_last_debug_frame): Remove deprecated debug-frame bits.
* libguile/stacks.c (scm_make_stack): Rework this function and its
dependents to only walk VM frames.
(scm_stack_id): Call out to the holder of the VM frame in question,
which should be a VM or a VM continuation, for the stack ID. Currently
this bit is stubbed out.
(scm_last_stack_frame): Removed. It seems this is mainly useful for a
debugger, and we need to rewrite the debugger to work on the Scheme
level.
* test-suite/tests/continuations.test ("continuations"): Remove test for
last-stack-frame.
* libguile/continuations.h (struct scm_t_contregs):
* libguile/continuations.c (scm_make_continuation):
(copy_stack_and_call, scm_i_with_continuation_barrier): No need to
save and restore debug frames.
* libguile/threads.h (scm_i_thread): Don't track debug frames.
(scm_i_last_debug_frame, scm_i_set_last_debug_frame): Remove macro
accessors.
* libguile/threads.c (guilify_self_1): Don't track debug frames.
* libguile/throw.c: No need to track debug frames in a jmpbuf.
* libguile/vm-engine.c (vm_engine, VM_PUSH_DEBUG_FRAMES): Don't push
debug frames.
* libguile/vm.h:
* libguile/vm.c (scm_vm_call_with_new_stack): New function. Currently
stubbed out though.
* libguile/frames.c, libguile/instructions.c, libguile/objcodes.c,
libguile/programs.c, libguile/throw.c, libguile/vm-i-scheme.c,
libguile/vm.c: Replace uses of discouraged constructs by their
current counterparts.
The crux of this problem was that the thread doing a throw, and so
checking scm_i_critical_section_level, was different from the thread
that was in a critical section.
* libguile/async.h (scm_i_critical_section_level): Removed, replaced
by per-thread critical_section_level.
(SCM_CRITICAL_SECTION_START, SCM_CRITICAL_SECTION_END): Use
per-thread critical_section_level.
* libguile/continuations.c (scm_dynthrow): Check per-thread
critical_section_level.
* libguile/threads.c (guilify_self_1): Init per-thread
critical_section_level.
(scm_i_critical_section_level): Removed.
* libguile/threads.h (scm_i_thread): New critical_section_level field.
* libguile/throw.c (scm_ithrow): Check per-thread critical_section_level.
* libguile/struct.c (scm_make_struct_layout, scm_struct_init)
(scm_struct_vtable_p, scm_struct_ref, scm_struct_set_x): use string
and symbol accessors and avoid unpacking strings and symbols
* libguile/throw.c (scm_ithrow): allow wide symbols in the error message
* libguile/unif.c (scm_enclose_array, scm_istr2bve): use string
accessors and avoid unpacking strings
Problem was that if an application includes both libguile.h and the
system's setjmp.h, and is compiled on IA64, it gets compile errors
because of jmp_buf, setjmp and longjmp being multiply defined.
* libguile/__scm.h (__ia64__): Define scm_i_jmp_buf, SCM_I_SETJMP and
SCM_I_LONGJMP instead of jmp_buf, setjmp and longjmp.
(all other platforms): Map scm_i_jmp_buf, SCM_I_SETJMP and
SCM_I_LONGJMP to jmp_buf, setjmp and longjmp.
* libguile/continuations.c (scm_make_continuation): Use `SCM_I_SETJMP'
instead of `setjmp'.
(copy_stack_and_call): Use `SCM_I_LONJMP' instead of `longjmp'.
(scm_ia64_longjmp): Use type `scm_i_jmp_buf' instead of `jmp_buf'.
* libguile/continuations.h (scm_t_contregs): Use type `scm_i_jmp_buf'
instead of `jmp_buf'.
* libguile/threads.c (suspend): Use `SCM_I_SETJMP' instead of
`setjmp'.
* libguile/threads.h (scm_i_thread): Use type `scm_i_jmp_buf' instead
of `jmp_buf'.
* libguile/throw.c (JBJMPBUF, make_jmpbuf, jmp_buf_and_retval): Use
type `scm_i_jmp_buf' instead of `jmp_buf'.
(scm_c_catch): Use `SCM_I_SETJMP' instead of `setjmp'.
(scm_ithrow): Use `SCM_I_LONGJMP' instead of `longjmp'.
* libguile/throw.c (scm_c_catch): Stash away the current vm's regs, and
restore them if there's a nonlocal exit. There is a terrible case we
have to handle if we catch from when the vm smob type isn't registered
but the throw has the vm registered, but I think we handle this fine.
* libguile/vm-engine.c (vm_run):
* libguile/vm-i-system.c (halt): Don't make a dynwind context, so that
entering the VM doesn't cons at all, except for the arg list. Maybe we
can fix that bit too.
* libguile/vm.c (vm_reset_stack): Remove, as there is no more dynwind.
(make_vm): Return #f if the tc16 hasn't yet been registered.
* Specific problems in IA64 make check
** test-unwind
Representation of the relevant dynamic context:
non-rewindable
catch frame make cont.
o----o-----a----------b-------------c
\
\ call cont.
o-----o-----------d
A continuation is captured at (c), with a non-rewindable frame in the
dynamic context at (b). If a rewind through that frame was attempted,
Guile would throw to the catch at (a). Then the context unwinds back
past (a), then winds forwards again, and the captured continuation is
called at (d).
We should end up at the catch at (a). On ia64, we get an "illegal
instruction".
The problem is that Guile does not restore the ia64 register backing
store (RBS) stack (which is saved off when the continuation is
captured) until all the unwinding and rewinding is done. Therefore,
when the rewind code (scm_i_dowinds) hits the non-rewindable frame at
(b), the RBS stack hasn't yet been restored. The throw finds the
jmp_buf (for the catch at (a)) correctly from the dynamic context, and
jumps back to (a), but the RBS stack is invalid, hence the illegal
instruction.
This could be fixed by restoring the RBS stack earlier, at the same
point (copy_stack) where the normal stack is restored. But that
causes a problem in the next test...
** continuations.test
The dynamic context diagram for this case is similar:
non-rewindable
catch frame make cont.
a----x-----o----------b-------------c
\
\ call cont.
o-------d
The only significant difference is that the catch point (a) is
upstream of where the dynamic context forks. This means that the RBS
stack at (d) already contains the correct RBS contents for throwing
back to (a), so it doesn't matter whether the RBS stack that was saved
off with the continuation gets restored.
This test passes with the Guile 1.8.4 code, but fails (with an
"illegal instruction") when the code is changed to restore the RBS
stack earlier as described above.
The problem now is that the RBS stack is being restored _too_ early;
specifically when there is still stuff to do that relies on the old
RBS contents. When a continuation is called, the sequence of relevant
events is:
(1) Grow the (normal) stack until it is bigger than the (normal)
stack saved off in the continuation. (scm_dynthrow, grow_stack)
(2) scm_i_dowinds calls itself recursively, such that
(2.1) for each rewind (from (x) to (c)) that will be needed,
another frame is added to the stack (both normal and RBS),
with local variables specifying the required rewind; the
rewinds don't actually happen yet, they will happen when
the stack unwinds again through these frames
(2.2) required unwinds - back from where the continuation was
called (d) to the fork point (x) - are done immediately.
(3) The normal (i.e. non-RBS) stack that was stored in the
continuation is restored (i.e. copied on top of the actual
stack).
Note that this doesn't overwrite the frames that were added in
(2.1), because the growth in (1) ensures that the added frames
are beyond the end of the restored stack.
(4) ? Restore the RBS stack here too ?
(5) Return (from copy_stack) through the (2.1) frames, which means
that the rewinds now happen.
(6) setcontext (or longjmp) to the context (c) where the
continuation was captured.
The trouble is that step (1) does not create space in the RBS stack in
the same kind of way that it does for the normal stack. Therefore, if
the saved (in the continuation) RBS stack is big enough, it can
overwrite the RBS of the (2.1) frames that still need to complete.
This causes an illegal instruction when we return through those frames
and try to perform the rewinds.
* Fix
The key to the fix is that the saved RBS stack only needs to be
restored at some point before the next setcontext call, and that doing
it as close to the setcontext call as possible will avoid bad
interactions with the pre-setcontext stack. Therefore we do the
restoration at the last possible point, immediately before the next
setcontext call.
The situation is complicated by there being two ways that the next
setcontext call can happen.
- If the unwinding and rewinding is all successful, the next
setcontext will be the one from step (6) above. This is the
"normal" continuation invocation case.
- If one of the rewinds throws an error, the next setcontext will
come from the throw implementation code. (And the one in step (6)
will never happen.) This is the rewind error case.
In the rewind error case, the code calling setcontext knows nothing
about the continuation. So to cover both cases, we:
- copy (in step (4) above) the address and length of the
continuation's saved RBS stack to the current thread state
(SCM_I_CURRENT_THREAD)
- modify all setcontext callers so that they check the current
thread state for a saved RBS stack, and restore it if so before
calling setcontext.
* Notes
** I think rewinders cannot rely on using any stack data
Unless it can be guaranteed that the data won't go into a register.
I'm not 100% sure about this, but I think it follows from the fact
that the RBS stack is not restored until after the rewinds have
happened.
Note that this isn't a regression caused by the current fix. In Guile
1.8.4, the RBS stack was restored _after_ the rewinds, and this is
still the case now.
** Most setcontext calls for `throw' don't need to change the RBS stack
In the absence of continuation invocation, the setcontext call in the
throw implementation code always sets context to a place higher up the
same stack (both normal and RBS), hence no stack restoration is
needed.
* Other changes
** Using setcontext for all non-local jumps (for __ia64__)
Along the way, I read a claim somewhere that setcontext was more
reliable than longjmp, in cases where the stack has been manipulated.
I don't now have any reason to believe this, but it seems reasonable
anyway to leave the __ia64__ code using getcontext/setcontext, instead
of setjmp/longjmp.
(I think the only possible argument against this would be performance -
if getcontext was significantly slower than setjmp. It that proves to
be the case, we should revisit this.)
** Capping RBS base for non-main threads
Somewhere else along the way, I hit a problem in GC, involving the RBS
stack of a non-main thread. The problem was, in
SCM_MARK_BACKING_STORE, that scm_ia64_register_backing_store_base was
returning a value that was massively greater than the value of
scm_ia64_ar_bsp, leading to a seg fault. This is because the
implementation of scm_ia64_register_backing_store_base is only valid
for the main thread. I couldn't find a neat way of getting the true
RBS base of a non-main thread, but one idea is simply to call
scm_ia64_ar_bsp when guilifying a thread, and use the value returned
as an upper bound for that thread's RBS base. (Note that the RBS
stack grows upwards.)
(Were it not for scm_init_guile, we could be much more definitive
about this. We could take the value of scm_ia64_ar_bsp as a
definitive base address for the part of the RBS stack that Guile cares
about. We could also then discard
scm_ia64_register_backing_store_base.)
scm_set_thread_cleanup_x, scm_thread_cleanup): Lock on thread-specific
admin mutex instead of `thread_admin_mutex'.
* threads.h (scm_i_thread)[admin_mutex]: New field.
* throw.c (make_jmpbuf): Don't enter critical section during thread
spawn -- there is a possibility of deadlock if other threads are
exiting.
