headers
J Chapman Flack | 1 Mar 2005 05:29
Picon
Favicon

Re: sun jdk 1.5 appletviewer deadlocks or crashes on 2.0 MP - am I crazy?

Christos Zoulas wrote:

> >  CPU:total 1 family 6, cmov, cx8, fxsr, mmx, sse

> It is getting it from procfs, and I will commit a fix for it shortly.

That nailed the problem - thanks!  I merged the miscfs/procfs/procfs_linux
and arch/i386/i386/procfs_machdep changes into my 2.0 sources and now
sun-jre15 runs applets just fine, both in the standalone appletviewer and
as the mozilla plugin.

I'm not sure if you intended the line    len = sizeof buf;
to remain in procfs_docpuinfo, though;
I removed it here.

Thanks again,
-Chap
Permalink | Reply | 0 comments |
headers
Jochen Kunz | 1 Mar 2005 10:11
Picon

Re: yamt-km branch

On Sun, 27 Feb 2005 21:47:18 +0900
Izumi Tsutsui <tsutsui <at> ceres.dti.ne.jp> wrote:

> Anyone who can test on 020/030 models?
If "everything fails" I can dig out a HP9000 400t. 68030, 50 MHz.
I think there is also a Mac IIci somewhere. 68030, 25 MHz.
Ahh, and there is a Sun 3/60. 68020, 20MHz
--

-- 

tschüß,
       Jochen

Homepage: http://www.unixag-kl.fh-kl.de/~jkunz/
Permalink | Reply | 0 comments |
headers
Chuck Silvers | 1 Mar 2005 10:53
Favicon

Re: direct I/O

On Mon, Feb 28, 2005 at 08:59:54AM -0400, Jared D.McNeill wrote:
> Solaris also supports 'forcedirectio' as a mount option. Are you 
> planning on supporting similar functionality as well?

we could do that too, it wouldn't be hard at all once we have O_DIRECT.

-Chuck
Permalink | Reply | 2 comments |
headers
Jaromir Dolecek | 1 Mar 2005 11:02
Picon

Re: direct I/O

On Tue, Mar 01, 2005 at 01:53:39AM -0800, Chuck Silvers wrote:
> On Mon, Feb 28, 2005 at 08:59:54AM -0400, Jared D.McNeill wrote:
> > Solaris also supports 'forcedirectio' as a mount option. Are you 
> > planning on supporting similar functionality as well?
> 
> we could do that too, it wouldn't be hard at all once we have O_DIRECT.

BTW, isn't O_DIRECT orthogonal to the loan-for-read()/write(), i.e.
wouldn't it be useful to have both? The loan-for-X only avoids
cost of copyin(), right?

Jaromir
--

-- 
Jaromir Dolecek <jdolecek <at> NetBSD.org>            http://www.NetBSD.cz/
-=- We can walk our road together if our goals are all the same;     -=-
-=- We can run alone and free if we pursue a different aim.          -=-
Permalink | Reply | 1 comment |
headers
Chuck Silvers | 1 Mar 2005 11:08
Favicon

Re: loaning for read() of regular files

On Mon, Feb 28, 2005 at 01:08:03PM -0800, Jason Thorpe wrote:
> 
> On Feb 27, 2005, at 11:00 PM, Chuck Silvers wrote:
> 
> >(instead I'm looking at direct I/O, which I'll post more about 
> >shortly.)
> 
> FreeBSD has some code for this.  Can we leverage it at all?  Last time 
> I looked at it, it looked fairly well isolated in the FFS code.

my version is fs-independent (except for the hook in the VOP_READ / VOP_WRITE
code to call it), and that seems more desirable than making it fs-specific.

-Chuck
Permalink | Reply | 1 comment |
headers
Chuck Silvers | 1 Mar 2005 11:30
Favicon

Re: direct I/O

On Mon, Feb 28, 2005 at 02:24:35PM -0800, Jonathan Stone wrote:
> 
> In message <20050228081036.GA20462 <at> spathi.chuq.com>,
> Chuck Silvers writes:
> 
> >hi folks,
> >
> >since my profiling experiments with the 10M Rows mysql benchmark showed that
> >avoiding read-before-overwrite didn't help much at all, this morning I decided
> >to go for the real solution and look into unbuffered (aka direct) I/O.
> >implementing this turned out to be much easier than I expected, it only
> >took a few hours to get it working.  the diff is at:
> >
> >	ftp://ftp.netbsd.org/pub/NetBSD/misc/chs/diff.directio
> 
> I hate to seem nagging, and I do appreciate the performance regression
> tests you did with loanout.
> 
> But: if FreeBSD-4 doesn't need direct I/O, why do we need it?  Could

it's not a question of need, I'm saying that direct I/o should perform
better than buffered I/O (even with loaning) for database-style applications
that do their own caching.

> the poor performance be symptomatic of something else -- e.g., by
> analogy to the relative NFS performance of NetBSD-2.0 and Free-BSD4,
> maybe we just lack sufficient read-ahead?

there appear to be two problems with the netbsd in the default configuration
in the 10M Rows benchmark:
(Continue reading)

Permalink | Reply | 10 comments |
headers
Chuck Silvers | 1 Mar 2005 11:33
Favicon

Re: direct I/O

On Mon, Feb 28, 2005 at 03:33:17PM -0800, Jason Thorpe wrote:
> 
> On Feb 28, 2005, at 2:24 PM, Jonathan Stone wrote:
> 
> >But: if FreeBSD-4 doesn't need direct I/O, why do we need it?
> 
> FreeBSD *has* direct I/O.  Perhaps MySQL is using it.

mysql can use direct I/O, but it doesn't by default (and it wasn't in
the configuration used in the article).

-Chuck
Permalink | Reply | 0 comments |
headers
Chuck Silvers | 1 Mar 2005 11:35
Favicon

Re: direct I/O

On Tue, Mar 01, 2005 at 11:02:38AM +0100, Jaromir Dolecek wrote:
> On Tue, Mar 01, 2005 at 01:53:39AM -0800, Chuck Silvers wrote:
> > On Mon, Feb 28, 2005 at 08:59:54AM -0400, Jared D.McNeill wrote:
> > > Solaris also supports 'forcedirectio' as a mount option. Are you 
> > > planning on supporting similar functionality as well?
> > 
> > we could do that too, it wouldn't be hard at all once we have O_DIRECT.
> 
> BTW, isn't O_DIRECT orthogonal to the loan-for-read()/write(), i.e.
> wouldn't it be useful to have both? The loan-for-X only avoids
> cost of copyin(), right?

yes, they're orthogonal.  I'm just thinking I'd rather implement
direct I/O first.

-Chuck
Permalink | Reply | 0 comments |
headers
Chuck Silvers | 1 Mar 2005 12:15
Favicon

Re: direct I/O

On Tue, Mar 01, 2005 at 08:48:08AM +1100, matthew green wrote:
> 
> 
> wow, cool.
> 
>    this is safer than the loaning-for-read()/write() stuff, since applications
>    must explicitly ask for the new behaviour and the interface to do this is
>    fairly standard already.  the interface is just a new open/fcntl flag,
>    O_DIRECT, which is a hint that I/O done via this file descriptor be performed
> 
> 
> WRT implementation, is this irix-alike or solaris-alike?  ie, for irix
> it is a per-fd thing, but for solaris, it is a per-vnode thing (ie, 
> turing on O_DIRECT for a file does it for all users, not just this fd.)

I did it per-fd.  solaris doesn't have O_DIRECT as far as I can tell.

-Chuck
Permalink | Reply | 1 comment |
headers
SODA Noriyuki | 1 Mar 2005 12:30
Picon

Re: direct I/O

>>>>> On Tue, 1 Mar 2005 03:15:58 -0800, Chuck Silvers <chuq <at> chuq.com> said:

> I did it per-fd.  solaris doesn't have O_DIRECT as far as I can tell.

Yeah.
It has directio(3C), though.

>> NAME
>>      directio - provide advice to file system
>> 
>> SYNOPSIS
>>      #include <sys/types.h>
>>      #include <sys/fcntl.h>
>> 
>>      int directio(int fildes, int advice);
>> 
>> DESCRIPTION
... SNIP ...
>>      The advice argument is kept per file;  the  last  caller  of
>>      directio()  sets  the  advice for all applications using the
>>      file associated with fildes.
>> 
>>      Values for advice are defined in <sys/fcntl.h>.
>> 
>>      DIRECTIO_OFF
... SNIP ...
>>      DIRECTIO_ON
--
soda
(Continue reading)

Permalink | Reply | 0 comments |

Gmane