From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from stargate3.asicdesigners.com (stargate.chelsio.com [12.32.117.8]) by dpdk.org (Postfix) with ESMTP id 746082B92 for ; Wed, 1 Jun 2016 10:34:41 +0200 (CEST) Received: from localhost (scalar.blr.asicdesigners.com [10.193.185.94]) by stargate3.asicdesigners.com (8.13.8/8.13.8) with ESMTP id u518YcUA010023; Wed, 1 Jun 2016 01:34:39 -0700 Date: Wed, 1 Jun 2016 14:04:34 +0530 From: Rahul Lakkireddy To: Bruce Richardson Cc: "dev@dpdk.org" , Kumar A S , Nirranjan Kirubaharan Message-ID: <20160601083433.GA4016@chelsio.com> References: <99991ceaae3c8ead727fb43d78dd62670214e573.1462519635.git.rahul.lakkireddy@chelsio.com> <20160531162013.GA28620@bricha3-MOBL3> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160531162013.GA28620@bricha3-MOBL3> User-Agent: Mutt/1.5.24 (2015-08-30) Subject: Re: [dpdk-dev] [PATCH 1/5] pci: fix access to PCI config space in bsd X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 01 Jun 2016 08:34:41 -0000 Hi Bruce, On Tuesday, May 05/31/16, 2016 at 09:20:13 -0700, Bruce Richardson wrote: > On Fri, May 06, 2016 at 01:13:15PM +0530, Rahul Lakkireddy wrote: > > PCIOCREAD and PCIOCWRITE ioctls to read/write PCI config space fail > > with EPERM due to missing write permission. Fix by opening /dev/pci/ > > with O_RDWR instead. > > > > Fixes: 632b2d1deeed ("eal: provide functions to access PCI config") > > > > Signed-off-by: Rahul Lakkireddy > > Signed-off-by: Kumar Sanghvi > > --- > > lib/librte_eal/bsdapp/eal/eal_pci.c | 4 ++-- > > 1 file changed, 2 insertions(+), 2 deletions(-) > > > > diff --git a/lib/librte_eal/bsdapp/eal/eal_pci.c b/lib/librte_eal/bsdapp/eal/eal_pci.c > > index 2d16d78..82330be 100644 > > --- a/lib/librte_eal/bsdapp/eal/eal_pci.c > > +++ b/lib/librte_eal/bsdapp/eal/eal_pci.c > > @@ -422,7 +422,7 @@ int rte_eal_pci_read_config(const struct rte_pci_device *dev, > > goto error; > > } > > > > - fd = open("/dev/pci", O_RDONLY); > > + fd = open("/dev/pci", O_RDWR); > > if (fd < 0) { > > RTE_LOG(ERR, EAL, "%s(): error opening /dev/pci\n", __func__); > > goto error; > > @@ -466,7 +466,7 @@ int rte_eal_pci_write_config(const struct rte_pci_device *dev, > > > > memcpy(&pi.pi_data, buf, len); > > > > - fd = open("/dev/pci", O_RDONLY); > > + fd = open("/dev/pci", O_RDWR); > > if (fd < 0) { > > RTE_LOG(ERR, EAL, "%s(): error opening /dev/pci\n", __func__); > > goto error; > > -- > Does the read function as well as the write one need O_RDWR permissions? There > is also an ioctl in rte_eal_pci_scan which operates on a RDONLY file descriptor. > Does that need to be modified also? Yes, both PCIOCREAD and PCIOCWRITE ioctls seem to require write permission. Otherwise, pci_ioctl [1] seems to return EPERM. On the other hand, the PCIOCGETCONF ioctl used in rte_eal_pci_scan doesn't seem to require a write permission. So, it should be fine to leave it as RDONLY. [1] https://svnweb.freebsd.org/base/release/10.3.0/sys/dev/pci/pci_user.c?revision=297553&view=markup#l493 Thanks, Rahul