From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from us-smtp-2.mimecast.com ([205.139.110.61]:43747 "EHLO us-smtp-delivery-1.mimecast.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1729546AbgAOJ5y (ORCPT ); Wed, 15 Jan 2020 04:57:54 -0500 Subject: Re: [kvm-unit-tests PATCH v7 4/4] s390x: SCLP unit test References: <20200110184050.191506-1-imbrenda@linux.ibm.com> <20200110184050.191506-5-imbrenda@linux.ibm.com> <8d7fb5c4-9e2c-e28a-16c0-658afcc8178d@redhat.com> <20200113133325.417bf657@p-imbrenda> <1b86b00a-261e-3d8c-fa52-c30e67463ad5@redhat.com> <20200113135832.1c6d3bb8@p-imbrenda> <22b5ce6a-18af-edec-efc6-e03450faddf8@redhat.com> <20200113150504.3fd218d5@p-imbrenda> <3db7eaf7-6020-365b-c849-9961e483352e@redhat.com> <20200113162439.7ae81f84@p-imbrenda> <9f0bee07-28bc-8154-3c67-402c82da8f89@redhat.com> <20200113171715.7334c1be@p-imbrenda> From: David Hildenbrand Message-ID: <0842fa2c-62f7-025c-ab01-145ea24328a1@redhat.com> Date: Wed, 15 Jan 2020 10:57:48 +0100 MIME-Version: 1.0 In-Reply-To: <20200113171715.7334c1be@p-imbrenda> Content-Type: text/plain; charset=windows-1252 Content-Language: en-US Content-Transfer-Encoding: quoted-printable Sender: linux-s390-owner@vger.kernel.org List-ID: To: Claudio Imbrenda Cc: kvm@vger.kernel.org, linux-s390@vger.kernel.org, thuth@redhat.com, borntraeger@de.ibm.com, frankja@linux.ibm.com On 13.01.20 17:17, Claudio Imbrenda wrote: > On Mon, 13 Jan 2020 17:06:05 +0100 > David Hildenbrand wrote: >=20 > [...] >=20 >>>>> this would be solved by adding special logic to the pgm interrupt >>>>> handler (as we have discussed in your previous email) >>>>> =20 >>>> >>>> I see, so the issue should hold for all SCLP checks where we don't >>>> expect an exception ... hmmm =20 >>> =20 >>> which is why my wrapper in the unit test is so complicated :) >>> =20 >> >> so .... if we would implement my suggestion (if we get an exception >> on a servc instruction, clear sclp_busy) that code would get >> simplified as well? :) >=20 > sure, as I said, that can be done. The question is if we really want to > change something in the interrupt handler (shared by all s390x unit > tests) just for the benefit of this one unit test. >=20 > Also consider that the changes to the interrupt handler would not > necessarily be trivial. i.e. you need to check that the origin of the > pgm interrupt is a SERVC instruction, and then act accordingly. >=20 I suggest something like the following: diff --git a/lib/s390x/interrupt.c b/lib/s390x/interrupt.c index 05f30be..d762e83 100644 --- a/lib/s390x/interrupt.c +++ b/lib/s390x/interrupt.c @@ -106,10 +106,17 @@ static void fixup_pgm_int(void) =20 void handle_pgm_int(void) { - if (!pgm_int_expected) + if (!pgm_int_expected) { + /* + * If we get a PGM interrupt while having sclp_busy=3Dtru= e, we + * will loop forever. Just force sclp_busy=3Dfalse to mak= e + * progress here. + */ + sclp_handle_ext(); report_abort("Unexpected program interrupt: %d at %#lx, i= len %d\n", lc->pgm_int_code, lc->pgm_old_psw.addr, lc->pgm_int_id); + } =20 pgm_int_expected =3D false; fixup_pgm_int(); Then this test could become something like (not sure about cc handling) diff --git a/s390x/sclp.c b/s390x/sclp.c index 10f0809..81c5a76 100644 --- a/s390x/sclp.c +++ b/s390x/sclp.c @@ -396,25 +396,18 @@ out: static void test_instbits(void) { SCCBHeader *h =3D (SCCBHeader *)pagebuf; - int cc; =20 - expect_pgm_int(); sclp_mark_busy(); h->length =3D 8; sclp_setup_int(); =20 asm volatile( - " .insn rre,0xb2204200,%1,%2\n" /* servc %1,%2 *= / - " ipm %0\n" - " srl %0,28" - : "=3D&d" (cc) : "d" (valid_code), "a" (__pa(pagebuf)) + " .insn rre,0xb2204200,%0,%1\n" + :: "d" (valid_code), "a" (__pa(pagebuf)) : "cc", "memory"); - if (lc->pgm_int_code) { - sclp_handle_ext(); - cc =3D 1; - } else if (!cc) - sclp_wait_busy(); - report(cc =3D=3D 0, "Instruction format ignored bits"); + sclp_wait_busy(); + report(true, "Instruction format ignored bits"); } This works correctly. E.g., adding a "*((uint8_t *)-50ul) =3D 2;" after the sclp_setup_int(); - to quickly fake a PGM exception - makes the test abort correctly: FAIL sclp-1g (0 tests) FAIL sclp-3g (0 tests) ABORT: sclp: Unexpected program interrupt: 5 at 0x155e8, ilen 6 --=20 Thanks, David / dhildenb