https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #17 from H. Peter Anvin ---
So I am still confused by this.
It would seem that this really ought to be a very simple request, and that
adding compiler support for all these cases would impose a really large burden
on the gcc team (y
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #16 from H. Peter Anvin ---
Except there is no load or store anywhere (see the case on comment 12), so I
don't understand.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #15 from Andrew Pinski ---
Maybe what you want is really a builtin does the load/store and the ability to
output the register # that the address is formed into instead of this
inline-asm. This might be better than what you are asking
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #12 from H. Peter Anvin ---
Certainly. This is *not* only used by copy_*_user (or {get,put}_user for that
matter), here is an example from msr.h:
static inline unsigned long long native_read_msr_safe(unsigned int msr,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #14 from H. Peter Anvin ---
Note: comment 13 is not intended to be rhetorical but is a genuine question.
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #13 from H. Peter Anvin ---
When you say "should be done in an exceptional way", could you please clarify
what you mean? I'm not sure I follow you there? Are you saying we should be
asking for compiler support?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #10 from Andrew Pinski ---
Your example is not a full example on how you use _ASM_EXTABLE_TYPE_REG .
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #11 from Andrew Pinski ---
Plus your example is only for copy_{to,from}_user which itself is an
exceptional part of the kernel and should be done in an exceptional way in the
first place.
Do you have other examples that is not from
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #9 from H. Peter Anvin ---
Created attachment 59450
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=59450&action=edit
Proposed assembly header implementation
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #8 from H. Peter Anvin ---
Created attachment 59449
--> https://gcc.gnu.org/bugzilla/attachment.cgi?id=59449&action=edit
Current code
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #7 from H. Peter Anvin ---
I have included a concrete example from the Linux kernel (with other parts of
the code stripped for clarity.)
The file asm_header.s shows how it could be implemented as an assembly header.
As you can see,
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #6 from H. Peter Anvin ---
No idea what you mean with #asmoptions.
Using hacks in the Makefile is equivalent to having to do dependencies by hand
(keep in mind that these statements will generally be part of header files.) In
other
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #5 from Richard Biener ---
So instead of having the #asminclude in the C file you can arrange for it
in the Makefile by source specific rules?
That said, at what point would you request a #asmoptions directive?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #4 from H. Peter Anvin ---
On October 22, 2024 1:19:05 PM PDT, "pinskia at gcc dot gnu.org"
wrote:
>https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
>
>Andrew Pinski changed:
>
> What|Removed |A
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
Andrew Pinski changed:
What|Removed |Added
Severity|normal |enhancement
Keywords|
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #3 from Andrew Pinski ---
Let me expand on this and question why what you want is needed.
Can you explain exactly what kind of macros that are asm macros are being used
and why they can't be C processor macros that will then be used
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
Andrew Pinski changed:
What|Removed |Added
Last reconfirmed||2024-10-22
Status|UNCONFIRM
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117265
--- Comment #1 from Andrew Pinski ---
Why not use the c preprocessor and string concating?
18 matches
Mail list logo
