Date: Sat, 18 Sep 1999 22:29:59 -0400 (EDT) From: "John W. DeBoskey" <jwd@unx.sas.com> To: wes@softweyr.com (Wes Peters) Cc: freebsd-hackers@freebsd.org Subject: Re: kern/13075 (was: Re: aio_*) Message-ID: <199909190229.WAA32777@bb01f39.unx.sas.com> In-Reply-To: From Wes Peters at "Sep 16, 1999 5:52: 1 pm"
next in thread | raw e-mail | index | archive | help
Hand up... I have two machines running heavily hit aio based file server mechanisms. If the patches will apply to a -current system about 2 weeks old I'll give it a try... (atleast without too much trouble). And now for a wish: /*----------------------------------------------------------+ | ST_AIO | | | | A task in the ST_AIO state means that one of our | | aio_writes has finished. we will loop thru all | | outstanding aio_writes to see which one completed. | | | *----------------------------------------------------------*/ case ST_AIO: /*-----------------------------------------------------+ | loop to get completed write process. | *-----------------------------------------------------*/ for (j=0; j<MAX_WRITERS; j++) { if (aio[j].task && aio_error(&aio[j].iocb) != EINPROGRESS) { ie: you have to loop to determine which aio operation completed. Looking thru the kernel, I don't see any easy way to fix this. But then, I may be missing something (which I don't understand) about how to setup the sigevent structure: Aio generates the signal for me (setup via the sigevent struct), but the sigev_value is not passed depending on how the kernel processes the signal. Thanks, John > Christopher Sedore wrote: > > > > On Thu, 16 Sep 1999, Wes Peters wrote: > > > > > Christopher Sedore wrote: > > > > > > > > On Wed, 15 Sep 1999, Ruslan Ermilov wrote: > > > > > > > > > > Do you by any change have an idea how to fix PR kern/13075 > > > > > (signal is not posted for async I/O on raw devices) > > > > > > > > Yes. There is no code to post the signal unless the job is of the lio > > > > variety. Writing a fix took about 15 minutes (including reboot). > > > > Extensive testing not included, though the test program provided with the > > > > PR now functions. > > > > > > Great, now do you want to tackle aio_cancel? ;^) > > > > I've been holding off to see whether I can get the other aio patches for > > improved socket io committed. I don't want to do two versions of > > aio_cancel (the changes for sockets alter the way things are queued and > > hence the way that cancels have to be done). > > Good to hear. OK, show of hands: who's using aio_* and has the time to > test patches for Christopher? > > - -- > "Where am I, and what am I doing in this handbasket?" > > Wes Peters Softweyr LLC > http://softweyr.com/ wes@softweyr.com > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-hackers" in the body of the message > > ------------------------------ > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199909190229.WAA32777>