Custom Query (1030 matches)
Results (694 - 696 of 1030)
| Ticket | Resolution | Summary | Owner | Reporter |
|---|---|---|---|---|
| #1238 | fixed | x86 assembler gets operand order wrong on MOVQ between two XMM registers | ||
| Description |
? (ccl::defx86lapfunction bug ()
(movq (% fp1) (% fp0))
(ret))
BUG
? (disassemble *)
L0
(leaq (@ (:^ L0) (% rip)) (% fn)) ; [0]
(movq (% fp0) (% fp1)) ; [7]
(retq) ; [11]
NIL
DISASSEMBLE is correct in this case. |
|||
| #1241 | fixed | Apparent bug causes data mangling | ||
| Description |
https://lists.clozure.com/pipermail/openmcl-devel/2014-November/010720.html |
|||
| #1242 | fixed | %get-errno x8632 (windows?) returns float | ||
| Description |
Set priority to major as this affects all stream reading. Running Windows 8.1 64bit though the code suggests my environment has little to do with the underlying issue. At REPL, this is the error,
There's only one declare statement in fd-input-available-p, see level-1\l1-streams.lisp lines 5552-5553, (let* ((res (int-errno-call (#_poll pollfds 1 (or milliseconds -1))))) (declare (fixnum res)) (int-errno-call) is a wrapped call to (%get-errno), see lib\macros.lisp lines 3912-3917 %get-errno, 32bit target, see level-0\X86\x86-io.lisp, is more complicated than I can parse. However, it seems pretty obvious that the code between line 43 (box-fixnum imm0 arg_z) and line 51 (single-value-return) is the issue. Verified that, on the same system, 64bit code functions fine (does not return a floating point). Obvious that this would be the case though after reviewing the 64bit implementation of %get-errno as there is no code between the box-fixnum and return. |
|||
