Three random questions

Three random questions

[Duraid Madina]

1) Has anyone pondered writing an HPUX-to-Linux ELF64 object 
demangler/translator? *damn* that aCC is good sometimes..

2) What is the Right Way to get huge pages at the moment? I'm running a 
2.6.5-rc and seem to be running across odd bugs using hugetlbfs.

3) Whatever happened to alloc_hugepages()? That strikes me as vastly (!) 
superior to hugetlbfs, at least for the scientist types who just want to 
escape horrible TLB thrashing.

     Curious,

     Duraid

RE: Three random questions

[Chen, Kenneth W]

Duraid Madina wrote on Monday, June 28, 2004 6:04 AM
> 2) What is the Right Way to get huge pages at the moment? I'm running a
> 2.6.5-rc and seem to be running across odd bugs using hugetlbfs.

Check out Documentation/vm/hugetlbpage.txt in the kernel source tree for
a detailed explanation.  If you found any bug in this area, please report
to the mailing list.  We fixed a few bugs last week. So might be good to
try latest kernel.


> 3) Whatever happened to alloc_hugepages()? That strikes me as vastly (!)
> superior to hugetlbfs, at least for the scientist types who just want to
> escape horrible TLB thrashing.

This was debated on the linux kernel mailing list extensively in the past,
check out the archive.

Re: Three random questions

[David Mosberger]

>>>>> On Mon, 28 Jun 2004 23:04:03 +1000, Duraid Madina <duraid@octopus.com.au> said:

  Duraid> 1) Has anyone pondered writing an HPUX-to-Linux ELF64 object
  Duraid> demangler/translator? *damn* that aCC is good sometimes..

It might be easier to generate assembly code.  I have done that
occasionally but the real issue is of course that HP-UX is big-endian,
so only endian-neutral code will work directly that way (and even so
there may be a licensing issue with running HP-UX generated code on a
Linux system; IANAL...).

	--david

Re: Three random questions

[Matthew Wilcox]

On Mon, Jun 28, 2004 at 11:16:21PM -0700, David Mosberger wrote:
> It might be easier to generate assembly code.  I have done that
> occasionally but the real issue is of course that HP-UX is big-endian,
> so only endian-neutral code will work directly that way (and even so
> there may be a licensing issue with running HP-UX generated code on a
> Linux system; IANAL...).

How could that be?  That's like the wackos who thought that compiling
your code with GCC would lead to you having to release all your source
code ...

-- 
"Next the statesmen will invent cheap lies, putting the blame upon 
the nation that is attacked, and every man will be glad of those
conscience-soothing falsities, and will diligently study them, and refuse
to examine any refutations of them; and thus he will by and by convince 
himself that the war is just, and will thank God for the better sleep 
he enjoys after this process of grotesque self-deception." -- Mark Twain