> -----Original Message-----
> From: Robert Latham [mailto:[EMAIL PROTECTED]
> Sent: Friday, May 26, 2006 1:50 PM
> To: Yan Lam
> Cc: [email protected]
> Subject: Re: [Pvfs2-users] pvfs2 on debian running on sparc hardware
> 
> On Fri, May 26, 2006 at 11:13:07AM -0700, Yan Lam wrote:
> > I'm having problem with pvfs2 running kernel 2.6.16 on sparc
> > architecture.  I complied the 1.4.0 source code and everything
> > compiled without any errors.  However when I load the kernel module,
> > it complaints about unknown symbol flush_ptrace_access and
> update_atime.
> 
> Hi
> 
> None of us have access to a sparc machine, so we don't get to test
this
> very often.  Thanks for reporting your problems.
> 
> the 'update_atime' problem we know about and have fixed in CVS.
> The symbol 'flush_ptrace_access', however, appears to be sparc64 only.
> 
> The 'cacheflush.h' header uses that symbol in two macros:
> copy_to_user_page and copy_from_user_page
> 
> It looks like the sparc kernel does not export the flush_ptrace_access
> symobl, even though its header files expose it to any routine that
> calls copy_to_user_page or copy_from_user_page.  I'd say this is a bug
> but can we work around it?
> 
> It turns out we can.  If you configure pvfs2 with the
--disable-kernel-
> aio option, then the PVFS2 kernel module will  use an alternate code
> path that might be a bit slower, but will not bring in
> any un-exported symbols.   You'll still have to work with CVS to get
> around the update_atime problem, but with these two changes you'll be
> on your way.

  I have recompiled the CVS source with --disable-kernel-aio.  When I
try to create the storage space, I get the error: "pvfs2-trove-dbpf:
//private/storage_attributes.db: No such file or directory"  The files
exist in the directory.  So, I bring up the server without the "-f"
option and no complaints.  However when I do a "pvfs2-ls /private", the
server crashed.

Here is the pvfs2-fs.conf and pvfs2server-localhost.conf
Pvfs2-fs.conf:
<Defaults>
        UnexpectedRequests 50
        LogFile /tmp/pvfs2-server.log
        EventLogging none
        LogStamp usec
        BMIModules bmi_tcp
        FlowModules flowproto_multiqueue
        PerfUpdateInterval 1000
        ServerJobBMITimeoutSecs 30
        ServerJobFlowTimeoutSecs 30
        ClientJobBMITimeoutSecs 300
        ClientJobFlowTimeoutSecs 300
        ClientRetryLimit 5
        ClientRetryDelayMilliSecs 2000
</Defaults>

<Aliases>
        Alias localhost tcp://localhost:3334
</Aliases>

<Filesystem>
        Name pvfs2-fs
        ID 2063684040
        RootHandle 1048576
        <MetaHandleRanges>
                Range localhost 4-2147483650
        </MetaHandleRanges>
        <DataHandleRanges>
                Range localhost 2147483651-4294967297
        </DataHandleRanges>
        <StorageHints>
                TroveSyncMeta yes
                TroveSyncData no
                AttrCacheKeywords datafile_handles,metafile_dist
                AttrCacheKeywords dir_ent, symlink_target
                AttrCacheSize 4093
                AttrCacheMaxNumElems 32768
        </StorageHints>
</Filesystem>

Pvfs2server.conf-localhost:
StorageSpace /private
HostID "tcp://localhost:3334"
LogFile /tmp/pvfs2-server.log


Any ideas.

Thanks,
Yan Lam - Systems Administrator
Unify Corporation
[EMAIL PROTECTED]

> 
> > Any idea on what I am missing?
> 
> You are missing nothing.  Thanks for the report, and we will update
our
> documentation.  Hopefully the sparc64 kernel guys will add
> flush_ptrace_access' to the list of exported symbols in the next
kernel
> release.
> 
> I should warn you that a few years ago someone trying to run pvfs2 on
a
> sparc III got a lot of SIGBUS errors when pvfs2 would carry out an
> unaligned memory access.   Since then, we have eliminated a lot of
> unalinged accesses, but there might still be a few lurking around.
> Let us know if you run into anything like that.
> 
> Thanks
> ==rob
> 
> --
> Rob Latham
> Mathematics and Computer Science Division    A215 0178 EA2D B059 8CDF
> Argonne National Labs, IL USA                B29D F333 664A 4280 315B



_______________________________________________
Pvfs2-users mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users

Reply via email to