Messages in this thread | | | Date | Wed, 1 Dec 1999 17:24:42 +0100 (MET) | From | Bjorn Wesen <> | Subject | Re: quickie on 2.0 ramdisk and buffers (solved) |
| |
I dug deeper and found that the culprit was ext2 in combination with the ramdisk. ext2 does bforget() instead of brelse() in ext2/truncate.c and bforget effectively nukes the buffer out of the ramdisk, so a new one has to be allocated next time the block is needed. It only happened when something needed truncation.
(I had a problem with that because of some customized buffer cache handling - I could just put the brelse instead and problem solved)
-Bjorn
On Wed, 1 Dec 1999, Bjorn Wesen wrote:
> I thought the idea with the ramdisk was that it is simply buffers that > don't belong to a physical device. When creating, say, a 512 kbyte > filesystem on /dev/ram0 the amount of buffers go up by about 512 kbyte so > that's ok. And if I write a file to the mounted ramdisk the amount of > buffers dont go up. But for example mkdir/rmdir causes yet more buffers to > be allocated. I found that weird. Before I go digging deep, I just wanted > to confirm that no more buffers should be allocated by the kernel for > operations on the ramdisk blocks ? Filesystem was ext2, kernel 2.0.33.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |