[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20070104094056.GI9608@in.ibm.com>
Date: Thu, 4 Jan 2007 15:10:56 +0530
From: Bharata B Rao <bharata@...ibm.com>
To: Suparna Bhattacharya <suparna@...ibm.com>
Cc: linux-aio@...ck.org, akpm@...l.org, drepper@...hat.com,
linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
jakub@...hat.com, mingo@...e.hu,
Sébastien Dugué <sebastien.dugue@...l.net>
Subject: [PATCHSET 3][PATCH 5/5][AIO] - Add listio support
listio support through a system call(lio_submit)
This builds on the infrastructure provided by Sebastien's AIO
completion signal notification and listio patches to provide listio
support via a new system call.
long lio_submit(aio_context_t ctx_id, int mode, long nr,
struct iocb __user * __user *iocbpp, struct sigevent __user *event)
More details about the system call appears within the patch itself.
Sebastien had taken the approach of overloading the io_submit() with
new aio_lio_opcode of IOCB_CMD_GROUP to support the listio behaviour.
And this is an alternative approach for supporting listio.
Would this system call approach be agreeable ?
Would it make sense to add another argument to support partial
completion notification ? i,e., generate a notification when a minimum
number of ios complete. Would such a feature be desirable ? Is it ok
to add one more argument to the system call for this purpose ?
This patch along with the previous 4 AIO completion signal notification
patches have been tested using libposix-aio library. Tests have been
done on x86 and x86_64 boxes. The compat syscall changes have been tested
on x86_64 system.
View attachment "aio-listio-support.patch" of type "text/plain" (16008 bytes)
Powered by blists - more mailing lists