diff options
| author | Bryan O'Sullivan <bos@pathscale.com> | 2006-07-01 04:36:15 -0700 |
|---|---|---|
| committer | Linus Torvalds <torvalds@g5.osdl.org> | 2006-07-01 09:56:01 -0700 |
| commit | 35783ec07c3f7f6902abe4433e7be1b664d0bbaf (patch) | |
| tree | f2bdf80a430862c46cc820b597fcb7cb88c7e777 /drivers/infiniband/hw/ipath/ipath_init_chip.c | |
| parent | 6d8e9dd050ea78862b6c5e2c873ad6407f9b2428 (diff) | |
[PATCH] IB/ipath: fix a bug that results in addresses near 0 being written via DMA
We can't tell for sure if any packets are in the infinipath receive buffer
when we shut down a chip port. Normally this is taken care of by orderly
shutdown, but when processes are terminated, or sending process has a bug, we
can continue to receive packets. So rather than writing zero to the address
registers for the closing port, we point it at a dummy memory.
Signed-off-by: Dave Olson <dave.olson@qlogic.com>
Signed-off-by: Bryan O'Sullivan <bryan.osullivan@qlogic.com>
Cc: "Michael S. Tsirkin" <mst@mellanox.co.il>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Diffstat (limited to 'drivers/infiniband/hw/ipath/ipath_init_chip.c')
| -rw-r--r-- | drivers/infiniband/hw/ipath/ipath_init_chip.c | 17 |
1 files changed, 17 insertions, 0 deletions
diff --git a/drivers/infiniband/hw/ipath/ipath_init_chip.c b/drivers/infiniband/hw/ipath/ipath_init_chip.c index 9faa201abe8e..ff322e5c14c8 100644 --- a/drivers/infiniband/hw/ipath/ipath_init_chip.c +++ b/drivers/infiniband/hw/ipath/ipath_init_chip.c @@ -647,6 +647,7 @@ int ipath_init_chip(struct ipath_devdata *dd, int reinit) u32 val32, kpiobufs; u64 val; struct ipath_portdata *pd = NULL; /* keep gcc4 happy */ + gfp_t gfp_flags = GFP_USER | __GFP_COMP; ret = init_housekeeping(dd, &pd, reinit); if (ret) @@ -834,6 +835,22 @@ int ipath_init_chip(struct ipath_devdata *dd, int reinit) else enable_chip(dd, pd, reinit); + + if (!ret && !reinit) { + /* used when we close a port, for DMA already in flight at close */ + dd->ipath_dummy_hdrq = dma_alloc_coherent( + &dd->pcidev->dev, pd->port_rcvhdrq_size, + &dd->ipath_dummy_hdrq_phys, + gfp_flags); + if (!dd->ipath_dummy_hdrq ) { + dev_info(&dd->pcidev->dev, + "Couldn't allocate 0x%lx bytes for dummy hdrq\n", + pd->port_rcvhdrq_size); + /* fallback to just 0'ing */ + dd->ipath_dummy_hdrq_phys = 0UL; + } + } + /* * cause retrigger of pending interrupts ignored during init, * even if we had errors |