| Message ID | 6b4bfba1-8c59-eb14-0307-1dcf828e1ca3@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Jiang, Apologies for the delayed follow-up on this. Comments below. On Sat, 2017-06-03 at 17:32 +0800, Jiang Yi wrote: > Hi Nic, > > If a se_cmd of COMPARE_AND_WRITE got aborted, "struct semaphore caw_sem" of the corresponding se_device will never being called up(). > > I propose a patch: > > > Signed-off-by: Jiang Yi <jiangyilism@gmail.com> > --- > drivers/target/target_core_transport.c | 3 +++ > 1 file changed, 3 insertions(+) > > diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c > index 6025935..02734ca 100644 > --- a/drivers/target/target_core_transport.c > +++ b/drivers/target/target_core_transport.c > @@ -726,6 +726,9 @@ void target_complete_cmd(struct se_cmd *cmd, u8 scsi_status) > */ > if (cmd->transport_state & CMD_T_ABORTED || > cmd->transport_state & CMD_T_STOP) { > + if ((cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE) && success) { > + up(&dev->caw_sem); > + } > spin_unlock_irqrestore(&cmd->t_state_lock, flags); > complete_all(&cmd->t_transport_stop_comp); > return; After thinking about this some more, releasing cam_sem here probably makes the most sense, instead of say invoking the completion callback based on se_cmd->transport_state, or invoking the completion callback from __transport_wait_for_tasks(). Although, the extra '&& success' check here is incorrect, as the releasing of cam_sem needs to happen for this special case regardless of which scsi_status that target_complete_cmd() is called with. So that said, applying this to target-pending/for-next, and dropping the extra check based on scsi_status. Thank you. -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c index 6025935..02734ca 100644 --- a/drivers/target/target_core_transport.c +++ b/drivers/target/target_core_transport.c @@ -726,6 +726,9 @@ void target_complete_cmd(struct se_cmd *cmd, u8 scsi_status) */ if (cmd->transport_state & CMD_T_ABORTED || cmd->transport_state & CMD_T_STOP) { + if ((cmd->se_cmd_flags & SCF_COMPARE_AND_WRITE) && success) { + up(&dev->caw_sem); + } spin_unlock_irqrestore(&cmd->t_state_lock, flags); complete_all(&cmd->t_transport_stop_comp); return;
Hi Nic, If a se_cmd of COMPARE_AND_WRITE got aborted, "struct semaphore caw_sem" of the corresponding se_device will never being called up(). I propose a patch: Signed-off-by: Jiang Yi <jiangyilism@gmail.com> --- drivers/target/target_core_transport.c | 3 +++ 1 file changed, 3 insertions(+)