NASM 2.05rc1
|
View:
New views
10 Messages
—
Rating Filter:
Alert me
|
 |
NASM 2.05rc1
by Frank Kotler-3
::
Rate this Message:
I don't think SF (or that drunken gnu mailman) likes me. This is a
retry. Feast or famine... H. Peter Anvin wrote: > Okay, it seems pretty clear that NASM 2.04 was a dud. Accordingly, I have tagged NASM 2.05rc1 to try to get a bug fix release out quickly. Please, people, try to: > > a) help test this so we don't end up in the same situation again, and > b) limit your checkins to bug fixes, or do them on a branch. Okay, got it pushed out to SF, and "advertised" a little... My own (fairly limited) testing indicates that the "main problem" (short jump out of range) is fixed. I'm still seeing warnings on "signed byte value exceeds bounds" when Nasm is deciding on its own initiative to optimize, for example "cmp eax, 0FFFFFFFFh", etc... This seems to crop up quite often (which means it's *doing* something for us!). Best, Frank ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by H. Peter Anvin
::
Rate this Message:
Frank Kotler wrote:
> > Okay, got it pushed out to SF, and "advertised" a little... > > My own (fairly limited) testing indicates that the "main problem" (short > jump out of range) is fixed. I'm still seeing warnings on "signed byte > value exceeds bounds" when Nasm is deciding on its own initiative to > optimize, for example "cmp eax, 0FFFFFFFFh", etc... This seems to crop > up quite often (which means it's *doing* something for us!). > Yes, I think that and the GOTPCREL issue should be looked at before 2.05. I will try to look at the above. I might have to call for help, though. -hpa ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by Charles Crayne
::
Rate this Message:
On Thu, 02 Oct 2008 23:27:26 -0700
"H. Peter Anvin" <hpa@...> wrote: > Yes, I think that and the GOTPCREL issue should be looked at before > 2.05 The warning is issued because 4294967295 (0FFFFFFFFh) is clearly greater than 127. And even when we coerce it to 0ffh, the usage deserves a warning message, because the code is not portable. If we really want to allow the users to express negative numbers in hex without the use of a minus sign, then we should allow size qualifiers preceding the constant, and default to the current bit size. Then we could fix the problem once, in the expression evaluator, instead of every place that the constant is used. -- Chuck http://www.pacificsites.com/~ccrayne/charles.html ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by H. Peter Anvin
::
Rate this Message:
Charles Crayne wrote:
> On Thu, 02 Oct 2008 23:27:26 -0700 > "H. Peter Anvin" <hpa@...> wrote: > >> Yes, I think that and the GOTPCREL issue should be looked at before >> 2.05 > > The warning is issued because 4294967295 (0FFFFFFFFh) is clearly > greater than 127. And even when we coerce it to 0ffh, the usage > deserves a warning message, because the code is not portable. > > If we really want to allow the users to express negative numbers in hex > without the use of a minus sign, then we should allow size qualifiers > preceding the constant, and default to the current bit size. Then we > could fix the problem once, in the expression evaluator, instead of > every place that the constant is used. > No, that's not correct. The usage is valid and portable, and is a valid optimization. If we *don't* make use of a valid optimization, then why do we have the optimizer in the first place? (Speaking as a programmer, I'd find an unpredictable optimizer worse than no optimizer at all.) The problem is that it is somewhat context dependent, and that the evaluator loses some information. Consider: add eax,0xffffffff ... can be validly optimized to add eax,byte -1 ... and ... and ax,0xffff ... can be validly converted to add ax,byte -1 ... but the trick, of course, is that add eax,0xffff is not optimizable, and one would logically expect a warning for add ax,0xffffffff Now, of course, the problem is that the evaluator doesn't tell us the difference between -128 and 0xffffffff_ffffff80, and I think it would be really sane to revamp it to make the distinction. The other part of this is probably to pay more direct attention to the flags to avoid stepping on cases where to optimizer might have been involved. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by Charles Crayne
::
Rate this Message:
On Thu, 02 Oct 2008 23:27:26 -0700
"H. Peter Anvin" <hpa@...> wrote: > GOTPCREL issue This is a carryover from the elf32 code, and what the message really means is that the support has not been implemented. You can assign this task to me. -- Chuck http://www.pacificsites.com/~ccrayne/charles.html ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by H. Peter Anvin
::
Rate this Message:
Charles Crayne wrote:
> On Thu, 02 Oct 2008 23:27:26 -0700 > "H. Peter Anvin" <hpa@...> wrote: > >> GOTPCREL issue > > This is a carryover from the elf32 code, and what the message really > means is that the support has not been implemented. You can assign this > task to me. > There isn't actually a bug report for it, but I'll note that you're handling it. I'll try to take on dealing with the warnings. -hpa ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by Frank Kotler-3
::
Rate this Message:
Frank Kotler wrote:
... > , and "advertised" a little... I guess I won't do *that* again! Just posted this to alt.lang.asm... as you can see it's got me taliking to myself... Frank Kotler wrote: > Frank Kotler wrote: >> Frank Kotler wrote: >>> Nasm 2.05rc1 is available at SourceForge. >>> >>> <http://sourceforge.net/project/showfiles.php?group_id=6208> >>> >>> Alternate (new!!!) address: >>> >>> http://www.nasm.us/ >> >> According to the download counters at SourceForge, downloads have fallen off sharply since I posted this. WTF??? I guess Nasm's millions of satisfied users are downloading from the new site so they can help us test it... > > Can't be that. 2.05rc1 is not yet available there. Maybe it's the economy... Yes it is! The "download" link just hasn't been updated... http://www.nasm.us/pub/nasm/releasebuilds/2.05rc1/ Best, Frank ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by H. Peter Anvin
::
Rate this Message:
Frank Kotler wrote:
> > > > Can't be that. 2.05rc1 is not yet available there. Maybe it's the > economy... > > Yes it is! The "download" link just hasn't been updated... > > http://www.nasm.us/pub/nasm/releasebuilds/2.05rc1/ > I set it up so the Download link points to the latest formal release. I could also point it to the releasebuilds directory, if you think that would make more sense. -hpa ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by Frank Kotler-3
::
Rate this Message:
H. Peter Anvin wrote:
> Frank Kotler wrote: >> > >> > Can't be that. 2.05rc1 is not yet available there. Maybe it's the >> economy... >> >> Yes it is! The "download" link just hasn't been updated... >> >> http://www.nasm.us/pub/nasm/releasebuilds/2.05rc1/ >> > > I set it up so the Download link points to the latest formal release. I > could also point it to the releasebuilds directory, if you think that > would make more sense. At this particular point, the latest formal release is gonna cause problems, I'm afraid. My fault for not getting the "candidates" out for testing. As a "general rule"... I still think the releasebuilds directory might be a good bet. Best, Frank ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
|
|
Re: NASM 2.05rc1
by H. Peter Anvin
::
Rate this Message:
Frank Kotler wrote:
> > At this particular point, the latest formal release is gonna cause > problems, I'm afraid. My fault for not getting the "candidates" out for > testing. As a "general rule"... I still think the releasebuilds > directory might be a good bet. > Only if you *don't* use the optimizer. The problem is in backwards compatibility mode (no -O option.) -hpa ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/ _______________________________________________ Nasm-devel mailing list Nasm-devel@... https://lists.sourceforge.net/lists/listinfo/nasm-devel |
| Free Forum Powered by Nabble | Forum Help |
