Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2004-07-22 18:17:23 +0000
\ Florin Cazacu:
>
> I did the following tests:
>
> I compiled a kernel with high memory support off. The drbd-0.7.0 is
> clean (i mean i didn't apply the patch for "zero copy network io").
>
> I get the errors on the primary when i use xfs. Everything seems to
> work ok when i use ext3 with blocksize of 1024.
> Now i'm applying the patch.... i get the same errors when i try to
> copy something on the drbd partition and the machines are connected.
so please revert that change for now,
and disable all use of drbd_send_page,
like below.
lge
===================================================================
--- drbd_main.c (revision 1447)
+++ drbd_main.c (working copy)
@@ -883,6 +883,7 @@
that we do not reuse our own buffer pages (EEs) to early, therefore
we have the net_ee list.
*/
+#if 0
int _drbd_send_page(drbd_dev *mdev, struct page *page,
int offset, size_t size)
{
@@ -923,6 +924,16 @@
mdev->send_cnt += size>>9;
return ok;
}
+#else
+int _drbd_send_page(drbd_dev *mdev, struct page *page,
+ int offset, size_t size)
+{
+ int ret;
+ ret = drbd_send(mdev, mdev->data.socket, kmap(page) + offset, size, 0);
+ kunmap(page);
+ return ret;
+}
+#endif
// Used to send write requests: bh->b_rsector !!
int drbd_send_dblock(drbd_dev *mdev, drbd_request_t *req)