public inbox for kvm@vger.kernel.org
 help / color / mirror / Atom feed
From: Christian Borntraeger <borntraeger@linux.ibm.com>
To: Janosch Frank <frankja@linux.ibm.com>, KVM <kvm@vger.kernel.org>
Cc: David Hildenbrand <david@kernel.org>,
	linux-s390 <linux-s390@vger.kernel.org>,
	Thomas Huth <thuth@redhat.com>,
	Claudio Imbrenda <imbrenda@linux.ibm.com>,
	Heiko Carstens <hca@linux.ibm.com>,
	Vasily Gorbik <gor@linux.ibm.com>,
	Alexander Gordeev <agordeev@linux.ibm.com>
Subject: Re: [PATCH v3] KVM: s390: vsie: Avoid injecting machine check on signal
Date: Mon, 16 Mar 2026 15:47:07 +0100	[thread overview]
Message-ID: <46e5720f-bd08-4076-bc10-e6834d58af73@linux.ibm.com> (raw)
In-Reply-To: <cf8b6d86-0919-4644-aaa3-de3241df2684@linux.ibm.com>



Am 16.03.26 um 14:57 schrieb Janosch Frank:
> On 3/16/26 14:09, Christian Borntraeger wrote:
>> The recent XFER_TO_GUEST_WORK change resulted in a situation, where the
>> vsie code would interpret a signal during work as a machine check during
>> SIE as both use the EINTR return code.
>> The exit_reason of the sie64a function has nothing to do with the
>> kvm_run exit_reason. Rename it and define a specific code for machine
>> checks instead of abusing -EINTR.
>> rename exit_reason into sie_return to avoid the naming conflict
>> and change the code flow in vsie.c to have a separate variable for rc
>> and sie_return.
>>
>> Fixes: 2bd1337a1295e ("KVM: s390: Use generic VIRT_XFER_TO_GUEST_WORK functions")
>> Signed-off-by: Christian Borntraeger <borntraeger@linux.ibm.com>
>> ---
>> v1->v2:
>> instead of a band-aid, make the machine check more direct
>> v2->v3:
>> avoid overloading the rc variable in vsie.c
>>
>>   arch/s390/include/asm/kvm_host.h   |  3 +++
>>   arch/s390/include/asm/stacktrace.h |  2 +-
>>   arch/s390/kernel/asm-offsets.c     |  2 +-
>>   arch/s390/kernel/entry.S           |  4 ++--
>>   arch/s390/kernel/nmi.c             |  4 ++--
>>   arch/s390/kvm/kvm-s390.c           | 15 ++++++++-------
>>   arch/s390/kvm/vsie.c               |  7 +++++--
>>   7 files changed, 22 insertions(+), 15 deletions(-)
>>
>> diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
>> index 64a50f0862aa..3039c88daa63 100644
>> --- a/arch/s390/include/asm/kvm_host.h
>> +++ b/arch/s390/include/asm/kvm_host.h
>> @@ -710,6 +710,9 @@ void kvm_arch_crypto_clear_masks(struct kvm *kvm);
>>   void kvm_arch_crypto_set_masks(struct kvm *kvm, unsigned long *apm,
>>                      unsigned long *aqm, unsigned long *adm);
>> +#define SIE64_RETURN_NORMAL    0
>> +#define SIE64_RETURN_MCCK    1
>> +
>>   int __sie64a(phys_addr_t sie_block_phys, struct kvm_s390_sie_block *sie_block, u64 *rsa,
>>            unsigned long gasce);
>> diff --git a/arch/s390/include/asm/stacktrace.h b/arch/s390/include/asm/stacktrace.h
>> index c9ae680a28af..ac3606c3babe 100644
>> --- a/arch/s390/include/asm/stacktrace.h
>> +++ b/arch/s390/include/asm/stacktrace.h
>> @@ -62,7 +62,7 @@ struct stack_frame {
>>           struct {
>>               unsigned long sie_control_block;
>>               unsigned long sie_savearea;
>> -            unsigned long sie_reason;
>> +            unsigned long sie_return;
>>               unsigned long sie_flags;
>>               unsigned long sie_control_block_phys;
>>               unsigned long sie_guest_asce;
>> diff --git a/arch/s390/kernel/asm-offsets.c b/arch/s390/kernel/asm-offsets.c
>> index e1a5b5b54e4f..fbd26f3e9f96 100644
>> --- a/arch/s390/kernel/asm-offsets.c
>> +++ b/arch/s390/kernel/asm-offsets.c
>> @@ -63,7 +63,7 @@ int main(void)
>>       OFFSET(__SF_EMPTY, stack_frame, empty[0]);
>>       OFFSET(__SF_SIE_CONTROL, stack_frame, sie_control_block);
>>       OFFSET(__SF_SIE_SAVEAREA, stack_frame, sie_savearea);
>> -    OFFSET(__SF_SIE_REASON, stack_frame, sie_reason);
>> +    OFFSET(__SF_SIE_RETURN, stack_frame, sie_return);
>>       OFFSET(__SF_SIE_FLAGS, stack_frame, sie_flags);
>>       OFFSET(__SF_SIE_CONTROL_PHYS, stack_frame, sie_control_block_phys);
>>       OFFSET(__SF_SIE_GUEST_ASCE, stack_frame, sie_guest_asce);
>> diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S
>> index 4873fe9d891b..5817cb47b2d0 100644
>> --- a/arch/s390/kernel/entry.S
>> +++ b/arch/s390/kernel/entry.S
>> @@ -200,7 +200,7 @@ SYM_FUNC_START(__sie64a)
>>       stg    %r3,__SF_SIE_CONTROL(%r15)    # ...and virtual addresses
>>       stg    %r4,__SF_SIE_SAVEAREA(%r15)    # save guest register save area
>>       stg    %r5,__SF_SIE_GUEST_ASCE(%r15)    # save guest asce
>> -    xc    __SF_SIE_REASON(8,%r15),__SF_SIE_REASON(%r15) # reason code = 0
>> +    xc    __SF_SIE_RETURN(8,%r15),__SF_SIE_RETURN(%r15) # return code = 0
>>       mvc    __SF_SIE_FLAGS(8,%r15),__TI_flags(%r14) # copy thread flags
>>       lmg    %r0,%r13,0(%r4)            # load guest gprs 0-13
>>       mvi    __TI_sie(%r14),1
>> @@ -237,7 +237,7 @@ SYM_INNER_LABEL(sie_exit, SYM_L_GLOBAL)
>>       xgr    %r4,%r4
>>       xgr    %r5,%r5
>>       lmg    %r6,%r14,__SF_GPRS(%r15)    # restore kernel registers
>> -    lg    %r2,__SF_SIE_REASON(%r15)    # return exit reason code
>> +    lg    %r2,__SF_SIE_RETURN(%r15)    # return sie return code
>>       BR_EX    %r14
>>   SYM_FUNC_END(__sie64a)
>>   EXPORT_SYMBOL(__sie64a)
>> diff --git a/arch/s390/kernel/nmi.c b/arch/s390/kernel/nmi.c
>> index a55abbf65333..fc1f775e4517 100644
>> --- a/arch/s390/kernel/nmi.c
>> +++ b/arch/s390/kernel/nmi.c
>> @@ -487,8 +487,8 @@ void notrace s390_do_machine_check(struct pt_regs *regs)
>>       mcck_dam_code = (mci.val & MCIC_SUBCLASS_MASK);
>>       if (test_cpu_flag(CIF_MCCK_GUEST) &&
>>       (mcck_dam_code & MCCK_CODE_NO_GUEST) != mcck_dam_code) {
>> -        /* Set exit reason code for host's later handling */
>> -        *((long *)(regs->gprs[15] + __SF_SIE_REASON)) = -EINTR;
>> +        /* Set sie return code for host's later handling */
>> +        *((long *)(regs->gprs[15] + __SF_SIE_RETURN)) = SIE64_RETURN_MCCK;
> 
> We don't need to cast this to long anymore, do we?
> Since sie_return is ulong, and we if we stay with positive numbers ulong would make the most sense.
> 
> Although I'd prefer a cast to stack_frame but I'm unsure if that would make it more readable if we stay with a one line change.
something like
-               *((long *)(regs->gprs[15] + __SF_SIE_RETURN)) = SIE64_RETURN_MCCK;
+               ((struct stack_frame *) regs->gprs[15])->sie_return = SIE64_RETURN_MCCK;

What do you think?

This reminds me that I should probably also make the local variable in vsie.c unsigned long instead of int.



  reply	other threads:[~2026-03-16 14:47 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-03-16 13:09 [PATCH v3] KVM: s390: vsie: Avoid injecting machine check on signal Christian Borntraeger
2026-03-16 13:13 ` Heiko Carstens
2026-03-16 13:42 ` Claudio Imbrenda
2026-03-16 13:57 ` Janosch Frank
2026-03-16 14:47   ` Christian Borntraeger [this message]
2026-03-16 15:11     ` Janosch Frank

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=46e5720f-bd08-4076-bc10-e6834d58af73@linux.ibm.com \
    --to=borntraeger@linux.ibm.com \
    --cc=agordeev@linux.ibm.com \
    --cc=david@kernel.org \
    --cc=frankja@linux.ibm.com \
    --cc=gor@linux.ibm.com \
    --cc=hca@linux.ibm.com \
    --cc=imbrenda@linux.ibm.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-s390@vger.kernel.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