Commits

Show all
Author Commit Message Labels Comments Date
Armin Rigo
Probably implement support for the GIL. It's just a few lines of code, but it would take serious efforts to write a test for it :-( Instead I suppose we can just give up and write a test for the whole pypy-c.
Branches
jitypes2
Armin Rigo
merge heads
Branches
jitypes2
Armin Rigo
Implement closing and reopening the stack.
Branches
jitypes2
Armin Rigo
Kill the two unnecessary copies of rsi and rdi at the start of the function.
Branches
jitypes2
Antonio Cuni
fix the test now that we generate a call_release_gil instead of call_may_force
Branches
jitypes2
Armin Rigo
Phew. Finally got a failing test. The libffi.Func needs to be promoted, which is done from the code in pypy/module/_ffi, but was not done automatically for direct usage.
Branches
jitypes2
Armin Rigo
Write the test by making sure that all opts are enabled. The test still passes, due to another issue (next checkin). Additionally, fix the fact that OS_LIBFFI_CALL calls can have random side-effects on random fields through callbacks.
Branches
jitypes2
Armin Rigo
merge heads
Branches
jitypes2
Armin Rigo
Add test_compile_framework_close_stack. In-progress. I don't understand so far how it passes, given that the functionality is not implemented.
Branches
jitypes2
Armin Rigo
A test for CALL_RELEASE_GIL with a callback.
Branches
jitypes2
Armin Rigo
Temporary implementation of CALL_RELEASE_GIL in x86, not actually releasing the gil at all nor closing the gc stack so far.
Branches
jitypes2
Antonio Cuni
fix the last occurence of types.void_p
Branches
jitypes2
Antonio Cuni
merge heads
Branches
jitypes2
Antonio Cuni
s/types.pointer/types.void_p
Branches
jitypes2
Antonio Cuni
rename _ffi.types.pointer to void_p, to underline that it is a untyped pointer
Branches
jitypes2
Antonio Cuni
conversion from arbitrary pointers to void* is always allowed
Branches
jitypes2
Antonio Cuni
the previous checkin was partly broken; now c_char_p and c_wchar_p are strong typed as well; there is still one failing test because now we are too strict and do not allow implicit conversions from typed pointers to void*
Branches
jitypes2
Antonio Cuni
check that the pointer types are compatible, before converting
Branches
jitypes2
Antonio Cuni
add a method to get the ffi type from the ctype type
Branches
jitypes2
Antonio Cuni
fix the signature of _as_ffi_pointer_, which changed recently
Branches
jitypes2
Antonio Cuni
typo
Branches
jitypes2
Antonio Cuni
fix translation
Branches
jitypes2
Antonio Cuni
merge heads
Branches
jitypes2
Antonio Cuni
introduce 'typed pointers', which carry info on which type they are pointing to; also, pass the argtype to the _as_ffi_pointer_ method; this should allow _ctypes to do type check on pointers when converting arguments
Branches
jitypes2
Armin Rigo
On x86-64, put the correct dfi assembler macros to allow gdb to debug the function (and walk past it in the backtrace).
Branches
jitypes2
Armin Rigo
Imports clean-up.
Branches
jitypes2
Armin Rigo
Fix test_logger.py.
Branches
jitypes2
Armin Rigo
Instead of storing the memo on the single global instance of Logger, create a new instance of LogOperations for each loop that we want to log, and attach it to the logged loop. This allows us to print later an operation from the loop, without the need for the global 'memo' field which breaks test_free_object.
Branches
jitypes2
Armin Rigo
Fix test.
Branches
jitypes2
Antonio Cuni
fix test
Branches
jitypes2
  1. Prev
  2. Next