linux/drivers/firewire, branch v3.14

Merge branch 'for-3.14-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq

2014-03-08T19:51:13Z

Pull workqueue fix from Tejun Heo: "This pull request contains a workqueue usage fix for firewire. For quite a long time now, workqueue only treats two work items identical iff both their addresses and callbacks match. This is to avoid introducing false dependency through the work item being recycled while being executed. This changes non-reentrancy guarantee for the users of PREPARE[_DELAYED]_WORK() - if the function changes, reentrancy isn't guaranteed against the previous instance. Firewire depended on such nonreentrancy guarantee. This is fixed by doing the work item multiplexing from firewire proper while keeping the work function unchanged" * 'for-3.14-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq: firewire: don't use PREPARE_DELAYED_WORK

firewire: don't use PREPARE_DELAYED_WORK

2014-03-07T15:19:57Z

PREPARE_[DELAYED_]WORK() are being phased out. They have few users and a nasty surprise in terms of reentrancy guarantee as workqueue considers work items to be different if they don't have the same work function. firewire core-device and sbp2 have been been multiplexing work items with multiple work functions. Introduce fw_device_workfn() and sbp2_lu_workfn() which invoke fw_device->workfn and sbp2_logical_unit->workfn respectively and always use the two functions as the work functions and update the users to set the ->workfn fields instead of overriding work functions using PREPARE_DELAYED_WORK(). This fixes a variety of possible regressions since a2c1c57be8d9 "workqueue: consider work function when searching for busy work items" due to which fw_workqueue lost its required non-reentrancy property. Signed-off-by: Tejun Heo Acked-by: Stefan Richter Cc: linux1394-devel@lists.sourceforge.net Cc: stable@vger.kernel.org # v3.9+ Cc: stable@vger.kernel.org # v3.8.2+ Cc: stable@vger.kernel.org # v3.4.60+ Cc: stable@vger.kernel.org # v3.2.40+

firewire: ohci: fix probe failure with Agere/LSI controllers

2014-03-06T20:36:23Z

Since commit bd972688eb24 "firewire: ohci: Fix 'failed to read phy reg' on FW643 rev8", there is a high chance that firewire-ohci fails to initialize LSI née Agere controllers. https://bugzilla.kernel.org/show_bug.cgi?id=65151 Peter Hurley points out the reason: IEEE 1394a:2000 clause 5A.1 (or IEEE 1394:2008 clause 17.2.1) say: "The PHY shall insure that no more than 10 ms elapse from the reassertion of LPS until the interface is reset. The link shall not assert LReq until the reset is complete." In other words, the link needs to give the PHY at least 10 ms to get the interface operational. With just the msleep(1) in bd972688eb24, the first read_phy_reg() during ohci_enable() may happen before the phy-link interface reset was finished, and fail. Due to the high variability of msleep(n) with small n, this failure was not fully reproducible, and not apparent at all with low CONFIG_HZ setting. On the other hand, Peter can no longer reproduce the issue with FW643 rev8. The read phy reg failures that happened back then may have had an unrelated cause. So, just revert bd972688eb24, except for the valid comment on TSB82AA2 cards. Reported-by: Mikhail Gavrilov Reported-by: Jay Fenlason Reported-by: Clemens Ladisch Reported-by: Peter Hurley Cc: stable@vger.kernel.org # v3.10+ Signed-off-by: Stefan Richter

firewire: net: fix use after free

2014-02-28T10:02:51Z

Commit 8408dc1c14c1 "firewire: net: use dev_printk API" introduced a use-after-free in a failure path. fwnet_transmit_packet_failed(ptask) may free ptask, then the dev_err() call dereferenced it. The fix is straightforward; simply reorder the two calls. Reported-by: Dan Carpenter Cc: stable@vger.kernel.org # v3.4+ Signed-off-by: Stefan Richter

firewire: Enable remote DMA above 4 GB

2014-01-20T00:11:13Z

This makes all of a machine's memory accessible to remote debugging via FireWire, using the physical response unit (i.e. RDMA) of OHCI-1394 link layer controllers. This requires actual support by the controller. The only ones currently known to support it are Agere/LSI FW643. Most if not all other OHCI-1394 controllers do not implement the optional Physical Upper Bound register. With them, RDMA will continue to be limited to the lowermost 4 GB. firewire-ohci's startup message in the kernel log is augmented to tell whether the controller does expose more than 4 GB to RDMA. While OHCI-1394 allows for a maximum Physical Upper Bound of 0xffff'0000'0000 (near 256 TB), this implementation sets it to 0x8000'0000'0000 (128 TB) in order to avoid interference with applications that require interrupt-served asynchronous request reception at respectively low addresses. Note, this change does not switch remote DMA on. It only increases the range of remote access to all memory (instead of just 4 GB) whenever remote DMA was switched on by other means. The latter is achieved by setting firewire-ohci's remote_dma parameter, or if the physical DMA filter is opened through firewire-sbp2. Derived from patch "firewire: Enable physical DMA above 4GB" by Peter Hurley from March 27, 2013. Signed-off-by: Stefan Richter

firewire: ohci: Turn remote DMA support into a module parameter

2014-01-12T17:54:38Z

This makes it possible to debug kernel over FireWire without the need to recompile it. [Stefan R: changed description from "...0" to "...N"] Cc: Dave Hansen Signed-off-by: Lubomir Rintel Signed-off-by: Stefan Richter

firewire: sbp2: bring back WRITE SAME support

2013-12-15T15:32:32Z

Commit 54b2b50c20a6 "[SCSI] Disable WRITE SAME for RAID and virtual host adapter drivers" disabled WRITE SAME support for all SBP-2 attached targets. But as described in the changelog of commit b0ea5f19d3d8 "firewire: sbp2: allow WRITE SAME and REPORT SUPPORTED OPERATION CODES", it is not required to blacklist WRITE SAME. Bring the feature back by reverting the sbp2.c hunk of commit 54b2b50c20a6. Signed-off-by: Stefan Richter Cc: stable@kernel.org

[SCSI] Disable WRITE SAME for RAID and virtual host adapter drivers

2013-11-29T04:48:39Z

Some host adapters do not pass commands through to the target disk directly. Instead they provide an emulated target which may or may not accurately report its capabilities. In some cases the physical device characteristics are reported even when the host adapter is processing commands on the device's behalf. This can lead to adapter firmware hangs or excessive I/O errors. This patch disables WRITE SAME for devices connected to host adapters that provide an emulated target. Driver writers can disable WRITE SAME by setting the no_write_same flag in the host adapter template. [jejb: fix up rejections due to eh_deadline patch] Signed-off-by: Martin K. Petersen Cc: stable@kernel.org Signed-off-by: James Bottomley

tree-wide: use reinit_completion instead of INIT_COMPLETION

2013-11-15T00:32:21Z

Use this new function to make code more comprehensible, since we are reinitialzing the completion, not initializing. [akpm@linux-foundation.org: linux-next resyncs] Signed-off-by: Wolfram Sang Acked-by: Linus Walleij (personally at LCE13) Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

firewire: ohci: Fix deadlock at bus reset

2013-08-29T20:35:05Z

Put bus_reset_work into its own workqueue. By doing this, forward progress of bus_reset_work() is guaranteed if the work is switched over to a rescuer thread. Switching work to a rescuer thread happens if a new worker thread could not be allocated in certain time (MAYDAY_INITIAL_TIMEOUT, typically 10 ms). This might not be possible under high memory pressure or even on a heavily loaded embedded system running a slow serial console. The former deadlock occured in the following situation: The rescuer thread ran fw_device_init->read_config_rom->read_rom->fw_run_transaction. fw_run_transaction blocked waiting for the completion object. This completion object would have been completed in bus_reset_work, but this work was never executed in the rescuer thread due to its strictly sequential behaviour. [Stefan R.: Removed WQ_NON_REENTRANT flag from allocation because it is no longer needed in current kernels. Add it back if you backport to kernels older than 3.7, i.e. one which does not contain dbf2576e37da "workqueue: make all workqueues non-reentrant". Swapped order of destroy_workqueue and pci_unregister_driver.] Signed-off-by: Stephan Gatzka Signed-off-by: Stefan Richter