Coda File System

Re: Linux/Alpha port banter

From: Peter J. Braam <braam_at_cs.cmu.edu>
Date: Tue, 10 Feb 1998 20:54:20 -0500 (EST)
Jim,

That's great.  The messages in the lwp test are a bit worrying.  Why don't
you find out _why_ it is printing that things are taking too long?

When you are a little further along, I'll tell you how to test the kernel
module (without having to build Venus).  We have a "potemkin" venus.  It's
probably worth first fixing rvm and compiling into the tree.

- Peter -

On Tue, 10 Feb 1998, Jim Doyle wrote:

> Get Ready, the port to Linux/Alpha is well on its way....
> Next is RVM and RPC2.  Coda Porting is definitely a therapeutic break
> from my DCE work. Maybe once I hand the work off to Peter to help cleanup
> Venus for 64-bits, I can go port the Minicache VFS to Solaris, hows that
> for a deal?
> 
> - I've fixed the 64-bit problems in the Linux VFS module.  The module
>   now loads through insmod and I see this. This doesnt mean anything,
>   it could still panic when tickled further.
> 
> 	jrd_at_italpha1$ more /proc/filesystems 
> 	        ext2
> 	        msdos
> 	nodev   proc
> 	nodev   nfs
> 	        iso9660
> 	nodev   coda
> 
> This shows up in the messages log:
> 
> Feb 10 15:41:32 italpha1 kernel: Coda Kernel/User communications module 0.04
> 
> - I've ported the LWP library to Alpha. When I ported DCE Threads to AXP,
>   I got to learn AXP Assembler - this was MUCH easier to port than DCE
>   threads.  Of course, I found this little nasty:
> 
>  struct	 lwp_ctl {			/* LWP control structure */
>         int	processcnt;		/* number of lightweight processes */
>         char	*outersp;		/* outermost stack pointer */
>         PROCESS	outerpid;		/* process carved by Initialize */
>     	PROCESS	first, last;		/* ptrs to first and last pcbs */
> ==>     long	dsptchstack[800];	/* stack for dispatcher use only */
> 					/* jrd_at_bu.edu 02/10/1998: stack align*/
>   }
> 
> This was previously:	
> 	char	dsptchstack[800];	/* stack for dispatcher use only */
> Which is all wrong, because on an ALPHA this means that the stack is
> not guaranteed to be longword aligned.
> 
> - I run the LWP tests:
> 
> jrd_at_italpha1$ ./test 5
> [ 15:00:00 ] ***LWP IO MANAGER(0x2010ab10) took too much cpu 0 secs  34180 usecs
> [ 17:40:01 ] ***LWP Main Process(0x20109870) took too much cpu 0 secs 109375 usecs
> [ 21:17:37 ] ***LWP Main Process(0x20109870) took too much cpu 0 secs  16602 usecs
> [ 18:48:18 ] ***LWP Main Process(0x20109870) took too much cpu 0 secs  15625 usecs
> [ 02:56:52 ] ***LWP Main Process(0x20109870) took too much cpu 0 secs  15625 usecs
> [ 01:10:15 ] ***LWP Main Process(0x20109870) took too much cpu 0 secs  15625 usecs
> 10140 milliseconds for 5 MWaits (2028125.000000 usec per Mwait and Signal)
> 
> jrd_at_italpha1$ ./rw 5
> 
> *Readers & Writers*
> 
> [There will be 5 readers]
> [Support initialized]
> [Creating Readers...
>     [Reader 0]
>     [Reader 1]
>     [Reader 2]
>     [Reader 3]
>     [Reader 4]
> done]
>     [Creating Writer...
>     [Writer]
> [4: Mary had a little lamb,]
> [0: Its fleece was white as snow,]
> [1: And everywhere that Mary went,]
> [2: The lamb was sure to go]
> [3: Mary had a little lamb,]
> done]
> [4: Its fleece was white as snow,]
> [0: And everywhere that Mary went,]
> [1: The lamb was sure to go]
> [2: Mary had a little lamb,]
> [3: Its fleece was white as snow,]
> [4: And everywhere that Mary went,]
> [0: The lamb was sure to go]
> [1: Mary had a little lamb,]
> [2: Its fleece was white as snow,]
> [3: And everywhere that Mary went,]
> [4: The lamb was sure to go]
> [0: Mary had a little lamb,]
> [1: Its fleece was white as snow,]
> [2: And everywhere that Mary went,]
> [3: The lamb was sure to go]
> [4: Mary had a little lamb,]
> [0: Its fleece was white as snow,]
> [1: And everywhere that Mary went,]
> [2: The lamb was sure to go]
> [3: Mary had a little lamb,]
> [4: Its fleece was white as snow,]
> [0: And everywhere that Mary went,]
> [1: The lamb was sure to go]
> [2: Mary had a little lamb,]
> [3: Its fleece was white as snow,]
> [4: And everywhere that Mary went,]
> [0: The lamb was sure to go]
> [1: Mary had a little lamb,]
> [2: Its fleece was white as snow,]
> [3: And everywhere that Mary went,]
> [4: The lamb was sure to go]
> [0: Mary had a little lamb,]
> [1: Its fleece was white as snow,]
> [2: And everywhere that Mary went,]
> [3: The lamb was sure to go]
> 
> *Exiting*
> 
> 
> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> Jim Doyle                         Boston University   Information Technology
> Systems Analyst/Programmer        email: jrd_at_bu.edu   Distributed Systems
> 						      tel. (617)-353-8248
> -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-++--+-+-+-+-+-+-
> 
> 
Received on 1998-02-10 20:57:04