From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([209.51.188.92]:50539) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1h0sQm-0006JD-1f for qemu-devel@nongnu.org; Mon, 04 Mar 2019 13:35:17 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1h0sQk-0002fb-VI for qemu-devel@nongnu.org; Mon, 04 Mar 2019 13:35:15 -0500 Date: Mon, 4 Mar 2019 19:35:09 +0100 From: Cornelia Huck Message-ID: <20190304193509.5eae1b41.cohuck@redhat.com> In-Reply-To: <1551466776-29123-12-git-send-email-jjherne@linux.ibm.com> References: <1551466776-29123-1-git-send-email-jjherne@linux.ibm.com> <1551466776-29123-12-git-send-email-jjherne@linux.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH v3 11/16] s390-bios: cio error handling List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Jason J. Herne" Cc: qemu-devel@nongnu.org, qemu-s390x@nongnu.org, pasic@linux.ibm.com, alifm@linux.ibm.com, borntraeger@de.ibm.com On Fri, 1 Mar 2019 13:59:31 -0500 "Jason J. Herne" wrote: > Add verbose error output for when unexpected i/o errors happen. This eases the > burden of debugging and reporting i/o errors. No error information is printed > in the success case, here is an example of what is output on error: > > cio device error > ssid : 0x0000000000000000 > cssid : 0x0000000000000000 > sch_no: 0x0000000000000000 > > Interrupt Response Block Data: > Function Ctrl : [Start] > Activity Ctrl : [Start-Pending] > Status Ctrl : [Alert] [Primary] [Secondary] [Status-Pending] > Device Status : [Unit-Check] > Channel Status : > cpa=: 0x000000007f8d6038 > prev_ccw=: 0x0000000000000000 > this_ccw=: 0x0000000000000000 > Eckd Dasd Sense Data (fmt 32-bytes): > Sense Condition Flags : > Residual Count =: 0x0000000000000000 > Phys Drive ID =: 0x000000000000009e > low cyl address =: 0x0000000000000000 > head addr & hi cyl =: 0x0000000000000000 > format/message =: 0x0000000000000008 > fmt-dependent[0-7] =: 0x0000000000000004 > fmt-dependent[8-15]=: 0xe561282305082fff > prog action code =: 0x0000000000000016 > Configuration info =: 0x00000000000040e0 > mcode / hi-cyl =: 0x0000000000000000 > cyl & head addr [0]=: 0x0000000000000000 > cyl & head addr [1]=: 0x0000000000000000 > cyl & head addr [2]=: 0x0000000000000000 > > Signed-off-by: Jason J. Herne > --- > pc-bios/s390-ccw/cio.c | 230 ++++++++++++++++++++++++++++++++++++++++++++++++ > pc-bios/s390-ccw/libc.h | 11 +++ > 2 files changed, 241 insertions(+) > > diff --git a/pc-bios/s390-ccw/cio.c b/pc-bios/s390-ccw/cio.c > index e61cfd3..c528bbf 100644 > --- a/pc-bios/s390-ccw/cio.c > +++ b/pc-bios/s390-ccw/cio.c > @@ -82,6 +82,228 @@ static bool irb_error(Irb *irb) > return irb->scsw.dstat != (SCSW_DSTAT_DEVEND | SCSW_DSTAT_CHEND); > } > > +static void print_eckd_dasd_sense_data(SenseDataEckdDasd *sd) > +{ > + char msgline[512]; > + > + if (sd->config_info & 0x8000) { > + sclp_print("Eckd Dasd Sense Data (fmt 24-bytes):\n"); > + } else { > + sclp_print("Eckd Dasd Sense Data (fmt 32-bytes):\n"); > + } > + > + strcat(msgline, " Sense Condition Flags :"); > + if (sd->status[0] & SNS_STAT0_CMD_REJECT) { > + strcat(msgline, " [Cmd-Reject]"); > + } > + if (sd->status[0] & SNS_STAT0_INTERVENTION_REQ) { > + strcat(msgline, " [Intervention-Required]"); > + } > + if (sd->status[0] & SNS_STAT0_BUS_OUT_CHECK) { > + strcat(msgline, " [Bus-Out-Parity-Check]"); > + } > + if (sd->status[0] & SNS_STAT0_EQUIPMENT_CHECK) { > + strcat(msgline, " [Equipment-Check]"); > + } > + if (sd->status[0] & SNS_STAT0_DATA_CHECK) { > + strcat(msgline, " [Data-Check]"); I'm wondering whether it would make sense to factor the common bits out. Might be overkill, though. > + } > + if (sd->status[0] & SNS_STAT0_OVERRUN) { > + strcat(msgline, " [Overrun]"); > + } > + if (sd->status[0] & SNS_STAT0_INCOMPL_DOMAIN) { > + strcat(msgline, " [Incomplete-Domain]"); > + } (...) > + sclp_print("cio device error\n"); > + print_int(" ssid ", schid.ssid); > + print_int(" cssid ", schid.cssid); > + print_int(" sch_no", schid.sch_no); > + sclp_print("\n"); > + print_irb_err(&irb); Maybe do basic_sense + print sense data only if there's actually a unit check? (Also, I'm not sure if you can even do a basic_sense in case e.g. of unexpected busy.) > + basic_sense(schid, &sd, sizeof(sd)); > + print_eckd_dasd_sense_data(&sd); > rc = -1; > break; > }