From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:54504) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1duYh4-000360-6P for qemu-devel@nongnu.org; Wed, 20 Sep 2017 02:41:11 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1duYgz-0003zA-58 for qemu-devel@nongnu.org; Wed, 20 Sep 2017 02:41:10 -0400 Received: from mx0b-001b2d01.pphosted.com ([148.163.158.5]:38920 helo=mx0a-001b2d01.pphosted.com) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1duYgy-0003y3-Vm for qemu-devel@nongnu.org; Wed, 20 Sep 2017 02:41:05 -0400 Received: from pps.filterd (m0098417.ppops.net [127.0.0.1]) by mx0a-001b2d01.pphosted.com (8.16.0.21/8.16.0.21) with SMTP id v8K6dYux019938 for ; Wed, 20 Sep 2017 02:41:04 -0400 Received: from e17.ny.us.ibm.com (e17.ny.us.ibm.com [129.33.205.207]) by mx0a-001b2d01.pphosted.com with ESMTP id 2d3brrqpvs-1 (version=TLSv1.2 cipher=AES256-SHA bits=256 verify=NOT) for ; Wed, 20 Sep 2017 02:41:03 -0400 Received: from localhost by e17.ny.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Wed, 20 Sep 2017 02:41:03 -0400 Date: Wed, 20 Sep 2017 14:40:58 +0800 From: Dong Jia Shi References: <20170913115029.47626-1-pasic@linux.vnet.ibm.com> <20170913115029.47626-5-pasic@linux.vnet.ibm.com> <20170919055005.GE5274@bjsdjshi@linux.vnet.ibm.com> <20170919114859.3e2d895b.cohuck@redhat.com> <20170919125717.1ae0cd38.cohuck@redhat.com> <9872cee8-3503-8d52-c68f-d9c61143201c@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <9872cee8-3503-8d52-c68f-d9c61143201c@linux.vnet.ibm.com> Message-Id: <20170920064058.GE11080@bjsdjshi@linux.vnet.ibm.com> Subject: Re: [Qemu-devel] [PATCH v2 4/4] s390x/css: support ccw IDA List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Halil Pasic Cc: Cornelia Huck , Dong Jia Shi , Pierre Morel , qemu-devel@nongnu.org * Halil Pasic [2017-09-19 14:04:03 +0200]: I have no problem with the rest parts of the discussion in this thread. > > > On 09/19/2017 12:57 PM, Cornelia Huck wrote: > >>>>> +static inline int ida_read_next_idaw(CcwDataStream *cds) > >>>>> +{ > >>>>> + union {uint64_t fmt2; uint32_t fmt1; } idaw; > >>>> ^ > >>>> Nit. > >>>> > >> Maybe checkpatch wanted it this way. My memories are blurry. > > > > I'd just leave it like that, tbh. > > > >>>>> + bool is_fmt2 = cds->flags & CDS_F_C64; > >>>>> + int ret; > >>>>> + hwaddr idaw_addr; > >>>>> + > >>>>> + if (is_fmt2) { > >>>>> + idaw_addr = cds->cda_orig + sizeof(idaw.fmt2) * cds->at_idaw; > >>>>> + if (idaw_addr & 0x07) { > >>>>> + return -EINVAL; /* channel program check */ > >>>>> + } > >>>>> + ret = address_space_rw(&address_space_memory, idaw_addr, > >>>>> + MEMTXATTRS_UNSPECIFIED, (void *) &idaw.fmt2, > >>>>> + sizeof(idaw.fmt2), false); > >>>>> + cds->cda = be64_to_cpu(idaw.fmt2); > > > >>>>> + } else { > >>>>> + idaw_addr = cds->cda_orig + sizeof(idaw.fmt1) * cds->at_idaw; > >>>>> + if (idaw_addr & 0x03) { > >>>> ?: > >>>> (idaw_addr & 0x80000003) > >>> Yes. > >>> > >> I will double check this. Does not seem unreasonable but > >> double-checking is better. > > Please let me know. I think the architecture says that the bit must be > > zero, and that we may (...) generate a channel program check. > > My fault... This is the address of an IDAW, not the content (data address) in an IDAW. So what Halil pointed out is the right direction to go I think. I will review in the thread of the new version (v3). > > Not exactly. The more significant bits part of the check > depend on the ccw format. This needs to be done for both > idaw formats. I would need to introduce a new flag, or > access the SubchDev to do this properly. > > Architecturally we also need to check the data addresses > from which we read so we have nothing bigger than > (1 << 31) - 1 if we are working with format-1 idaws. Right. This is what I actually wanted to say. > > I also think we did not take proper care of proper > maximum data address checks prior CwwDataStream which also > depend on the ccw format (in absence of IDAW or MIDAW). > > The ccw format dependent maximum address checks are (1 << 24) - 1 > and (1 << 31) - 1 respectively for format-0 and format-1 (on > the first indirection level that is for non-IDA for the data, > and for (M)IDA for the (M)IDAWs). > > Reference: > PoP pages 16-25 and 16-26 "Invalid IDAW or MIDAW Addre" and > "Invalid Data Address". > > How shall we proceed? > > Halil > > >>>> > >>>>> + return -EINVAL; /* channel program check */ > >>>>> + > >>>>> + } > >>>>> + ret = address_space_rw(&address_space_memory, idaw_addr, > >>>>> + MEMTXATTRS_UNSPECIFIED, (void *) &idaw.fmt1, > >>>>> + sizeof(idaw.fmt1), false); > >>>>> + cds->cda = be64_to_cpu(idaw.fmt1);>>>>> + } > >>>>> + ++(cds->at_idaw); > >>>>> + if (ret != MEMTX_OK) { > >>>>> + /* assume inaccessible address */ > >>>>> + return -EINVAL; /* channel program check */ > >>>>> + > >>>>> + } > >>>>> + return 0; > >>>>> +} -- Dong Jia Shi