Re: master 52b67740d10: Generalise a LAP optimisation rule

[Joseph Mingrone]

On Tue, 2023-04-04 at 08:10, Andrea Corallo <akrl@sdf.org> wrote:

> Joseph Mingrone <jrm@ftfl.ca> writes:

>> On Sat, 2023-03-25 at 12:22, Mattias Engdegård <mattiase@acm.org> wrote:

>>> branch: master
>>> commit 52b67740d10df8ca539fdc2c7d50283997683141
>>> Author: Mattias Engdegård <mattiase@acm.org>
>>> Commit: Mattias Engdegård <mattiase@acm.org>

>>>     Generalise a LAP optimisation rule

>>>     * lisp/emacs-lisp/byte-opt.el (byte-optimize-lapcode):
>>>     Accept (stack-set 1) as equivalent to (discardN-preserve-tos 1) in a
>>>     rule previously overlooked.  This is usually beneficial in code size
>>>     and almost always shortens dynamic paths.
>>> ---
>>>  lisp/emacs-lisp/byte-opt.el | 4 +++-
>>>  1 file changed, 3 insertions(+), 1 deletion(-)

>>> diff --git a/lisp/emacs-lisp/byte-opt.el b/lisp/emacs-lisp/byte-opt.el
>>> index 3c7aeb89525..0891ec80beb 100644
>>> --- a/lisp/emacs-lisp/byte-opt.el
>>> +++ b/lisp/emacs-lisp/byte-opt.el
>>> @@ -2765,7 +2765,9 @@ If FOR-EFFECT is non-nil, the return value is assumed 
>>> to be of no importance."
>>>                        (or (memq (caar tmp) '(byte-discard byte-discardN))
>>>                            ;; Make sure we don't hoist a 
>>> discardN-preserve-tos
>>>                            ;; that really should be merged or deleted 
>>> instead.
>>> -                          (and (eq (caar tmp) 'byte-discardN-preserve-tos)
>>> +                          (and (or (eq (caar tmp) 
>>> 'byte-discardN-preserve-tos)
>>> +                                   (and (eq (caar tmp) 'byte-stack-set)
>>> +                                        (eql (cdar tmp) 1)))
>>>                                 (let ((next (cadr tmp)))
>>>                                   (not (or (memq (car next)
>>>                                                  
>>> '(byte-discardN-preserve-tos

>> Hello,

>> This commit causes native compilation builds on FreeBSD to fail with the 
>> error

>>      native-ice ("org/org-plot.el" "block does not end with a branch"...

>> The full error is below and a link to a complete build log (FreeBSD
>> package build) is also included.

>> When the commit is reverted, the build finishes successfully again.

>> Joe


> Hi Joe,

> thanks for reporting, yes we are tracking this as bug#62537.

> Best Regards

>   Andrea

Hi Andrea,

Sure and thanks to you and Eli for clarification.  Also, apologies for
not locating bug#62537 before reporting there.

Kind regards,

Joe