On Thu, 2006-12-21 at 08:42 +0100, Jens Axboe wrote: > On Wed, Dec 20 2006, Ming Zhang wrote: > > On Wed, 2006-12-20 at 14:09 -0600, Seetharami Seelam wrote: > > > > > > On 12/20/06, Ming Zhang <[EMAIL PROTECTED]> wrote: > > > On Wed, 2006-12-20 at 17:45 +0100, Jens Axboe wrote: > > > > On Wed, Dec 20 2006, Ming Zhang wrote: > > > > > > > > > > On Wed, 2006-12-20 at 14:02 +0100, Jens Axboe wrote: > > > > > > On Wed, Dec 20 2006, Jens Axboe wrote: > > > > > > > On Wed, Dec 20 2006, Jens Axboe wrote: > > > > > > > > Irk, do_pipe() needs to know what the input is, > > > obviously. Lemme add > > > > > > > > that as well. > > > > > > > > > > > > > > This has a better chance of working, still not tested > > > though. If you can > > > > > > > test, I'll commit it once we have it working. > > > > > > > > > > > > It works for me, just tested it. Patch committed. > > > > > > > > > > > > > > > > yes, works. thx. > > > > > > > > > > now the fifo will use a name pattern like foo. but regular > > > file will use > > > > > foo.blktrace.X and if you give full name, blkparse does > > > not report any > > > > > warning or error, just return 0 results. this drove me > > > nuts before i > > > > > read the code. > > > > > > > > > > [ [EMAIL PROTECTED] blktrace]$ ./blkparse a.a.blktrace.0 > > > > > WARNING: full file name given. Should give trace name foo, > > > > > instead of file name foo.blktrace.x > > > > > > > > Better would be to just fix it up for the user, the current > > > setup is a > > > > little un-intuitive. Care to patch that up? :-) > > > > > > > > > > are these what u want? > > > > > > * user supplied a foo, we automatically match it with > > > foo.blktrace.X and > > > open. > > > * user supplied a foo.blktrace.X, we do not add extra > > > blktace.X and open > > > it directly. > > > > > > then if user have foo.blktrace.0 and foo.blktrace.1, current > > > code works > > > when use "-i foo". then shall we support "-i foo.blktrace.0" > > > and open > > > foo.blktrace.1 automatically? > > > > > > > > > Perhaps, if user explicity supplies the name(s), you should > > > open just that (those) file(s). May be you should open all > > > when the name is supplied as foo.blktrace.* > > > > key point here is not the implementation difficulty, but how we decide a > > consistent rule. i agree with the rules u set, if i understand > > correctly. > > > > * if "-", then read from stdin; > > * if file name is foo.blktrace.*, then we try to open all; > > * all other file name pattern, we open only _one_ file with _exact_ file > > name. > > > > See if others like this. > > I think that is the best approach, if the case where the full name is > given we check and warn if other CPU files are there. It could just be a > pilot error, and we should warn in that case. > > "You specified file foo.blktrace.0 and files from other CPUs also exist. > blkparse will only read the given file, which may not be what you want. > Use 'foo' as the filename to read all saved data." > > or something to that effect.
ok. i will give it a try. > - To unsubscribe from this list: send the line "unsubscribe linux-btrace" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
