On Thu, Feb 20, 2014 at 2:30 AM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote: > B1;3202;0cOn Wed, 19 Feb 2014, Dan Williams wrote: > >> Since commit 77873803363c "net_dma: mark broken" we no longer pin dma >> engines active for the network-receive-offload use case. As a result >> the ->free_chan_resources() that occurs after the driver self-test no >> longer has a NET_DMA induced ->alloc_chan_resources() to back it up. A >> late firing irq can lead to ksoftirqd spinning indefinitely due to the >> tasklet_disable() performed by ->free_chan_resources(). Only >> ->alloc_chan_resources() can clear this condition in affected kernels. >> >> This problem has been present since commit 3e037454bcfa "I/OAT: Add >> support for MSI and MSI-X" in 2.6.24, but is now exposed. Given the >> NET_DMA use case is deprecated we can revisit moving the driver to use >> threaded irqs. For now, just tear down the irq and tasklet properly by: > > Right, moving to threaded irqs would get rid of the whole tasklet > mess. > >> 1/ Disable the irq from triggering the tasklet >> >> 2/ Disable the irq from re-arming >> >> 3/ Flush inflight interrupts >> >> 4/ Flush the timer >> >> 5/ Flush inflight tasklets >> >> References: >> https://lkml.org/lkml/2014/1/27/282 >> https://lkml.org/lkml/2014/2/19/672 >> >> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx> >> Cc: Ingo Molnar <mingo@xxxxxxx> >> Cc: Steven Rostedt <rostedt@xxxxxxxxxxx> >> Cc: <stable@xxxxxxxxxxxxxxx> >> Reported-by: Mike Galbraith <bitbucket@xxxxxxxxx> >> Reported-by: Stanislav Fomichev <stfomichev@xxxxxxxxxxxxxx> >> Signed-off-by: Dan Williams <dan.j.williams@xxxxxxxxx> > > Reviewed-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx> > > As Mike pointed out tsi721_free_chan_resources() has the same issue. > > I did a quick scan of all tasklet_disable() sites. The teardown or > similar wreckage is available in: > > drivers/atm/he.c > drivers/dma/at_hdmac.c > drivers/dma/pch_dma.c > drivers/input/keyboard/omap-keypad.c > drivers/isdn/gigaset/interface.c > drivers/media/pci/mantis/mantis_dvb.c > drivers/mmc/host/s3cmci.c > drivers/net/ethernet/jme.c > drivers/net/ethernet/silan/sc92031.c > drivers/net/usb/r8152.c > drivers/net/wireless/mwl8k.c > drivers/ntb/ntb_hw.c > drivers/rapidio/devices/tsi721_dma.c > drivers/s390/crypto/ap_bus.c > drivers/spi/spi-pl022.c > drivers/staging/cxt1e1/linux.c > drivers/staging/ozwpan/ozhcd.c > drivers/usb/gadget/fsl_qe_udc.c > > That's 18 of 30 usage sites. Impressive.... > > We need to poke the relevant maintainers to get this solved. > Maybe also rename tasklet_disable() to tasklet_pause() to make it clearer "this isn't the tasklet cleanup routine you're looking for"? -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
