Re: [patch] fix for pagecache fs corruption (2.3.15-pre2) [Re:

[Andrea Arcangeli]

On Mon, 23 Aug 1999, Linus Torvalds wrote:

>You're working around the problem, methinks.

I am not working around the problem, I am only working to fix the ramdisk
driver. Incidentally I did things handy also to the common case (and
necessary with the current inode allocation).

>It's not a hack, and it cannot be exploited. You can't just lock down
>pages in memory - they will be paged out on demand, and there is no

I can even lockdown cache-page pages in memory if I start using a ramdisk
device. So IMO it make no sense to enlarge a lot the icache just because
we are not able to convert the page-cache-overlapped-buffers to
regular-hashed-buffers. Also considering that we must be capable of
converting the page-cache-overlapped-buffers to regular-hashed-buffers
anyway if we want to be able to `cp /dev/ramdisk image' after we unmounted
the filesystem.

Andrea