IB/hfi1: Fix an Oops on pci device force remove

This patch fixes an Oops on device unbind, when the device is used
by a PSM user process. PSM processes access device resources which
are freed on device removal. Similar protection exists in uverbs
in ib_core for Verbs clients, but PSM doesn't use ib_uverbs hence
a separate protection is required for PSM clients.

Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Dean Luick <dean.luick@intel.com>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Tadeusz Struk <tadeusz.struk@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This commit is contained in:
Tadeusz Struk
2016-10-25 08:57:55 -07:00
committed by Doug Ledford
parent d9ac4555fb
commit acd7c8fe14
4 changed files with 44 additions and 5 deletions

View File

@ -1538,12 +1538,31 @@ bail:
return ret;
}
static void wait_for_clients(struct hfi1_devdata *dd)
{
/*
* Remove the device init value and complete the device if there is
* no clients or wait for active clients to finish.
*/
if (atomic_dec_and_test(&dd->user_refcount))
complete(&dd->user_comp);
wait_for_completion(&dd->user_comp);
}
static void remove_one(struct pci_dev *pdev)
{
struct hfi1_devdata *dd = pci_get_drvdata(pdev);
/* close debugfs files before ib unregister */
hfi1_dbg_ibdev_exit(&dd->verbs_dev);
/* remove the /dev hfi1 interface */
hfi1_device_remove(dd);
/* wait for existing user space clients to finish */
wait_for_clients(dd);
/* unregister from IB core */
hfi1_unregister_ib_device(dd);
@ -1558,8 +1577,6 @@ static void remove_one(struct pci_dev *pdev)
/* wait until all of our (qsfp) queue_work() calls complete */
flush_workqueue(ib_wq);
hfi1_device_remove(dd);
postinit_cleanup(dd);
}