qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Peter Jin <pjin@linux.ibm.com>
To: Eric Farman <farman@linux.ibm.com>,
	pasic@linux.ibm.com, borntraeger@linux.ibm.com,
	richard.henderson@linaro.org, david@redhat.com,
	cohuck@redhat.com, thuth@redhat.com, mjrosato@linux.ibm.com
Cc: peter@peterjin.org, qemu-s390x@nongnu.org, qemu-devel@nongnu.org
Subject: Re: [PATCH v2] s390x/css: revert SCSW ctrl/flag bits on error
Date: Fri, 28 Oct 2022 16:26:03 -0400	[thread overview]
Message-ID: <f1abceef-d85a-217d-09d1-88e7a023b48a@linux.ibm.com> (raw)
In-Reply-To: <3a265985495a653b0cb5b4a1cf2cfad29f734dc0.camel@linux.ibm.com>


On 10/28/22 16:22, Eric Farman wrote:
> On Thu, 2022-10-27 at 23:23 +0200, Peter Jin wrote:
>> Revert the control and flag bits in the subchannel status word in
>> case
>> the SSCH operation fails with non-zero CC (ditto for CSCH and HSCH).
>> According to POPS, the control and flag bits are only changed if
>> SSCH,
>> CSCH, and HSCH return CC 0, and no other action should be taken
>> otherwise.
>> In order to simulate that after the fact, the bits need to be
>> reverted on
>> non-zero CC.
>>
> I'm okay to this point...
>
>> This change is necessary due to the fact that the pwrite() in vfio-
>> ccw
>> which triggers the SSCH can fail at any time. Previously, there was
>> only virtio-ccw, whose do_subchannel_work function was only able to
>> return CC0. However, once vfio-ccw went into the mix, it has become
>> necessary to handle errors in code paths that were previously assumed
>> to always return success.
>>
>> In our case, we found that in case of pwrite() failure (which was
>> discovered by strace injection), the subchannel could be stuck in
>> start
>> pending state, which could be problematic if the pwrite() call
>> returns
>> CC2. Experimentation shows that the guest tries to retry the SSCH
>> call as
>> normal for CC2, but it actually continously fails due to the fact
>> that
>> the subchannel is stuck in start pending state even though no start
>> function is actually taking place.
> ...but the two paragraphs above are a bit cumbersome to digest. Maybe
> it's just too late in the week for me. What about something like this?
>
> """
> While the do_subchannel_work logic for virtual (virtio) devices will
> return condition code 0, passthrough (vfio) devices may encounter
> errors from either the host kernel or real hardware that need to be
> accounted for after this point. This includes restoring the state of
> the Subchannel Status Word to reflect the subchannel, as these bits
> would not be set in the event of a non-zero condition code from the
> affected instructions.
>
> Experimentation has shown that a failure on a START SUBCHANNEL (SSCH)
> to a passthrough device would leave the subchannel with the START
> PENDING activity control bit set, thus blocking subsequent SSCH
> operations in css_do_ssch() until some form of error recovery was
> undertaken since no interrupt would be expected.
> """
>
>> Signed-off-by: Peter Jin <pjin@linux.ibm.com>
> We've talked previously about clearing this within the
> do_subchannel_work_passthrough routine in order to keep the _virtual
> paths untouched, but this seems like a reasonable approach to me.
>
> The commit message is probably fine either way, but as far as the code
> goes:
>
> Reviewed-by: Eric Farman <farman@linux.ibm.com>
Okay, that sounds much better :)
>> ---
>>   hw/s390x/css.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++-
>> --
>>   1 file changed, 48 insertions(+), 3 deletions(-)
>>
>> diff --git a/hw/s390x/css.c b/hw/s390x/css.c
>> index 7d9523f811..95d1b3a3ce 100644
>> --- a/hw/s390x/css.c
>> +++ b/hw/s390x/css.c
>> @@ -1522,21 +1522,37 @@ IOInstEnding css_do_xsch(SubchDev *sch)
>>   IOInstEnding css_do_csch(SubchDev *sch)
>>   {
>>       SCHIB *schib = &sch->curr_status;
>> +    uint16_t old_scsw_ctrl;
>> +    IOInstEnding ccode;
>>   
>>       if (~(schib->pmcw.flags) & (PMCW_FLAGS_MASK_DNV |
>> PMCW_FLAGS_MASK_ENA)) {
>>           return IOINST_CC_NOT_OPERATIONAL;
>>       }
>>   
>> +    /*
>> +     * Save the current scsw.ctrl in case CSCH fails and we need
>> +     * to revert the scsw to the status quo ante.
>> +     */
>> +    old_scsw_ctrl = schib->scsw.ctrl;
>> +
>>       /* Trigger the clear function. */
>>       schib->scsw.ctrl &= ~(SCSW_CTRL_MASK_FCTL |
>> SCSW_CTRL_MASK_ACTL);
>>       schib->scsw.ctrl |= SCSW_FCTL_CLEAR_FUNC | SCSW_ACTL_CLEAR_PEND;
>>   
>> -    return do_subchannel_work(sch);
>> +    ccode = do_subchannel_work(sch);
>> +
>> +    if (ccode != IOINST_CC_EXPECTED) {
>> +        schib->scsw.ctrl = old_scsw_ctrl;
>> +    }
>> +
>> +    return ccode;
>>   }
>>   
>>   IOInstEnding css_do_hsch(SubchDev *sch)
>>   {
>>       SCHIB *schib = &sch->curr_status;
>> +    uint16_t old_scsw_ctrl;
>> +    IOInstEnding ccode;
>>   
>>       if (~(schib->pmcw.flags) & (PMCW_FLAGS_MASK_DNV |
>> PMCW_FLAGS_MASK_ENA)) {
>>           return IOINST_CC_NOT_OPERATIONAL;
>> @@ -1553,6 +1569,12 @@ IOInstEnding css_do_hsch(SubchDev *sch)
>>           return IOINST_CC_BUSY;
>>       }
>>   
>> +    /*
>> +     * Save the current scsw.ctrl in case HSCH fails and we need
>> +     * to revert the scsw to the status quo ante.
>> +     */
>> +    old_scsw_ctrl = schib->scsw.ctrl;
>> +
>>       /* Trigger the halt function. */
>>       schib->scsw.ctrl |= SCSW_FCTL_HALT_FUNC;
>>       schib->scsw.ctrl &= ~SCSW_FCTL_START_FUNC;
>> @@ -1564,7 +1586,13 @@ IOInstEnding css_do_hsch(SubchDev *sch)
>>       }
>>       schib->scsw.ctrl |= SCSW_ACTL_HALT_PEND;
>>   
>> -    return do_subchannel_work(sch);
>> +    ccode = do_subchannel_work(sch);
>> +
>> +    if (ccode != IOINST_CC_EXPECTED) {
>> +        schib->scsw.ctrl = old_scsw_ctrl;
>> +    }
>> +
>> +    return ccode;
>>   }
>>   
>>   static void css_update_chnmon(SubchDev *sch)
>> @@ -1605,6 +1633,8 @@ static void css_update_chnmon(SubchDev *sch)
>>   IOInstEnding css_do_ssch(SubchDev *sch, ORB *orb)
>>   {
>>       SCHIB *schib = &sch->curr_status;
>> +    uint16_t old_scsw_ctrl, old_scsw_flags;
>> +    IOInstEnding ccode;
>>   
>>       if (~(schib->pmcw.flags) & (PMCW_FLAGS_MASK_DNV |
>> PMCW_FLAGS_MASK_ENA)) {
>>           return IOINST_CC_NOT_OPERATIONAL;
>> @@ -1626,11 +1656,26 @@ IOInstEnding css_do_ssch(SubchDev *sch, ORB
>> *orb)
>>       }
>>       sch->orb = *orb;
>>       sch->channel_prog = orb->cpa;
>> +
>> +    /*
>> +     * Save the current scsw.ctrl and scsw.flags in case SSCH fails
>> and we need
>> +     * to revert the scsw to the status quo ante.
>> +     */
>> +    old_scsw_ctrl = schib->scsw.ctrl;
>> +    old_scsw_flags = schib->scsw.flags;
>> +
>>       /* Trigger the start function. */
>>       schib->scsw.ctrl |= (SCSW_FCTL_START_FUNC |
>> SCSW_ACTL_START_PEND);
>>       schib->scsw.flags &= ~SCSW_FLAGS_MASK_PNO;
>>   
>> -    return do_subchannel_work(sch);
>> +    ccode = do_subchannel_work(sch);
>> +
>> +    if (ccode != IOINST_CC_EXPECTED) {
>> +        schib->scsw.ctrl = old_scsw_ctrl;
>> +        schib->scsw.flags = old_scsw_flags;
>> +    }
>> +
>> +    return ccode;
>>   }
>>   
>>   static void copy_irb_to_guest(IRB *dest, const IRB *src, const PMCW
>> *pmcw,


  reply	other threads:[~2022-10-28 20:27 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-10-27 21:23 [PATCH v2] s390x/css: revert SCSW ctrl/flag bits on error Peter Jin
2022-10-28 20:22 ` Eric Farman
2022-10-28 20:26   ` Peter Jin [this message]
2022-10-31 14:08   ` Matthew Rosato
2022-10-31 14:44     ` Peter Jin
2022-11-06 10:13   ` Thomas Huth

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=f1abceef-d85a-217d-09d1-88e7a023b48a@linux.ibm.com \
    --to=pjin@linux.ibm.com \
    --cc=borntraeger@linux.ibm.com \
    --cc=cohuck@redhat.com \
    --cc=david@redhat.com \
    --cc=farman@linux.ibm.com \
    --cc=mjrosato@linux.ibm.com \
    --cc=pasic@linux.ibm.com \
    --cc=peter@peterjin.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-s390x@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=thuth@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).