lkml.org 
[lkml]   [2018]   [Apr]   [27]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v2 5/5] vfio: ccw: add traceponits for interesting error paths
On Mon, 23 Apr 2018 13:01:13 +0200
Dong Jia Shi <bjsdjshi@linux.vnet.ibm.com> wrote:

typo in subject: s/traceponits/tracepoints/

> From: Halil Pasic <pasic@linux.vnet.ibm.com>
>
> Add some tracepoints so we can inspect what is not working as is should.
>
> Signed-off-by: Halil Pasic <pasic@linux.vnet.ibm.com>
> Signed-off-by: Dong Jia Shi <bjsdjshi@linux.vnet.ibm.com>
> ---
> drivers/s390/cio/Makefile | 1 +
> drivers/s390/cio/vfio_ccw_fsm.c | 16 +++++++-
> drivers/s390/cio/vfio_ccw_trace.h | 77 +++++++++++++++++++++++++++++++++++++++
> 3 files changed, 93 insertions(+), 1 deletion(-)
> create mode 100644 drivers/s390/cio/vfio_ccw_trace.h


> @@ -135,6 +142,8 @@ static void fsm_io_request(struct vfio_ccw_private *private,
> goto err_out;
>
> io_region->ret_code = cp_prefetch(&private->cp);
> + trace_vfio_ccw_cp_prefetch(get_schid(private),
> + io_region->ret_code);
> if (io_region->ret_code) {
> cp_free(&private->cp);
> goto err_out;
> @@ -142,11 +151,13 @@ static void fsm_io_request(struct vfio_ccw_private *private,
>
> /* Start channel program and wait for I/O interrupt. */
> io_region->ret_code = fsm_io_helper(private);
> + trace_vfio_ccw_fsm_io_helper(get_schid(private),
> + io_region->ret_code);
> if (io_region->ret_code) {
> cp_free(&private->cp);
> goto err_out;
> }
> - return;
> + goto out;
> } else if (scsw->cmd.fctl & SCSW_FCTL_HALT_FUNC) {
> /* XXX: Handle halt. */
> io_region->ret_code = -EOPNOTSUPP;
> @@ -159,6 +170,9 @@ static void fsm_io_request(struct vfio_ccw_private *private,
>
> err_out:
> private->state = VFIO_CCW_STATE_IDLE;
> +out:
> + trace_vfio_ccw_io_fctl(scsw->cmd.fctl, get_schid(private),
> + io_region->ret_code);
> }
>
> /*

I really don't want to bikeshed, especially as some tracepoints are
better than no tracepoints, but...

We now trace fctl/schid/ret_code unconditionally (good).

We trace the outcome of cp_prefetch() and fsm_io_helper()
unconditionally. We don't, however, trace all things that may go wrong.
We have the tracepoint at the end, but it cannot tell us where the
error came from. Should we have tracepoints in every place (in this
function) that may generate an error? Only if there is an actual error?
Are the two enough for common debug scenarios?

Opinions? We can just go ahead with this and improve things later on, I
guess.

\
 
 \ /
  Last update: 2018-04-27 12:14    [W:0.105 / U:1.072 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site