Date: Thu, 17 Feb 2005 15:18:53 -0600 From: Eric Anderson <anderson@centtech.com> To: Chuck Swiger <cswiger@mac.com> Cc: current@freebsd.org Subject: Re: where did all my memory go? (file system cache) Message-ID: <42150A3D.8080500@centtech.com> In-Reply-To: <4214FD67.7060801@mac.com> References: <1108584730.95661.12.camel@server.mcneil.com> <20050216201716.GA28436@odin.ac.hmc.edu> <4213B3C8.3090508@centtech.com> <1108588393.12275.9.camel@server.mcneil.com> <20050216214031.GA2787@odin.ac.hmc.edu> <4213D3AA.70809@elischer.org> <84dead72050217081540fd7640@mail.gmail.com> <4214FD67.7060801@mac.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Chuck Swiger wrote:
> Joseph Koshy wrote:
>
>>> what I want is:
>>>
>>> int fd = open("myfile",...);
>>> write1GBfiletodisk(fd, data);
>>> ioctl(fd, PURGEFROMCACHE);
>>> perform_md5(fd);
>>>
>>> and be sure that teh MD5 is that of what is on the disk.
>>> not what is in RAM.
>>
>>
>> unmount(file-system-of("myfile")) (even if it fails) ?
>
>
> That's actually a pretty good suggestion, and is less intrusive than,
> say rebooting, which is probably the only way to be entirely sure that
> the write cache on the drive itself has been flushed. If the write
> cache is off, Julian probably ought to be able to trust fsync(2)...?
Wouldn't there be a way to take the code that does the cache dumping (excuse my bad lingo here) and make a little tool that does it without any actual unmounting? Suppose the filesystem actually unmounted.. yikes!
Eric
--
------------------------------------------------------------------------
Eric Anderson Sr. Systems Administrator Centaur Technology
I have seen the future and it is just like the present, only longer.
------------------------------------------------------------------------
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?42150A3D.8080500>