From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from wes1-so2.wedos.net (wes1-so2.wedos.net [46.28.106.16]) by dpdk.org (Postfix) with ESMTP id D6D5D37A6 for ; Wed, 10 Feb 2016 13:18:43 +0100 (CET) Received: from pcviktorin.fit.vutbr.cz (pcviktorin.fit.vutbr.cz [147.229.13.147]) by wes1-so2.wedos.net (Postfix) with ESMTPSA id 3q0g9b4YvJz6pL; Wed, 10 Feb 2016 13:18:43 +0100 (CET) Date: Wed, 10 Feb 2016 13:20:06 +0100 From: Jan Viktorin To: David Marchand Message-ID: <20160210132006.56f10e48@pcviktorin.fit.vutbr.cz> In-Reply-To: References: <1454076516-21591-1-git-send-email-david.marchand@6wind.com> <1454076516-21591-9-git-send-email-david.marchand@6wind.com> <20160210122333.276ee305@pcviktorin.fit.vutbr.cz> Organization: RehiveTech MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: "dev@dpdk.org" Subject: Re: [dpdk-dev] [PATCH 8/9] pci: add a helper to refresh a device 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, 10 Feb 2016 12:18:44 -0000 On Wed, 10 Feb 2016 13:00:50 +0100 David Marchand wrote: > On Wed, Feb 10, 2016 at 12:23 PM, Jan Viktorin wrote: > > On Fri, 29 Jan 2016 15:08:35 +0100 > > David Marchand wrote: > > > >> It will be used mainly for hotplug code. > >> > >> Signed-off-by: David Marchand > >> --- > >> lib/librte_eal/bsdapp/eal/eal_pci.c | 49 +++++++++++++++++++++++++++++++++++ > >> lib/librte_eal/common/eal_private.h | 13 ++++++++++ > >> lib/librte_eal/linuxapp/eal/eal_pci.c | 13 ++++++++++ > >> 3 files changed, 75 insertions(+) > >> > >> diff --git a/lib/librte_eal/bsdapp/eal/eal_pci.c b/lib/librte_eal/bsdapp/eal/eal_pci.c > >> index 4584522..5dd89e3 100644 > >> --- a/lib/librte_eal/bsdapp/eal/eal_pci.c > >> +++ b/lib/librte_eal/bsdapp/eal/eal_pci.c > >> @@ -396,6 +396,55 @@ error: > >> return -1; > >> } > >> > >> +int > >> +pci_refresh_device(const struct rte_pci_addr *addr) > > > > What about pci_reload_device or pci_reload_device_info? I don't mind > > too much, only the word 'refresh' reminds me other associations. > > Or maybe pci_update_device ? > I added pci_add_device in my other pci patchset, so update sounds better to me. > OK. > > >> +{ > >> + int fd; > >> + struct pci_conf matches[2]; > >> + struct pci_match_conf match = { > >> + .pc_sel = { > >> + .pc_domain = addr->domain, > >> + .pc_bus = addr->bus, > >> + .pc_dev = addr->devid, > >> + .pc_func = addr->function, > >> + }, > >> + }; > >> + struct pci_conf_io conf_io = { > >> + .pat_buf_len = 0, > >> + .num_patterns = 1, > >> + .patterns = { &match }, > >> + .match_buf_len = sizeof(matches), > >> + .matches = &matches[0], > >> + }; > >> + > >> + fd = open("/dev/pci", O_RDONLY); > > > > Just courious who provides this special file... is a DPDK-specific > > thing? I haven't noticed it anywhere in Linux. > > I don't know, just took the bsd pci code and plugged myself in it. > So for me this is a special bsd device. > > This is mainly copy/paste. > Look at rte_eal_pci_scan() from lib/librte_eal/bsdapp/eal/eal_pci.c. BSD... I didn't notice. That's the answer. > > > > >> + if (fd < 0) { > >> + RTE_LOG(ERR, EAL, "%s(): error opening /dev/pci\n", __func__); > >> + goto error; > > > > If you write: > > return -1; > > > > then you can... > > > >> + } > >> + > >> + if (ioctl(fd, PCIOCGETCONF, &conf_io) < 0) { > >> + RTE_LOG(ERR, EAL, "%s(): error with ioctl on /dev/pci: %s\n", > >> + __func__, strerror(errno)); > >> + goto error; > >> + } > >> + > >> + if (conf_io.num_matches != 1) > >> + goto error; > >> + > >> + if (pci_scan_one(fd, &matches[0]) < 0) > >> + goto error; > >> + > >> + close(fd); > >> + > >> + return 0; > >> + > >> +error: > > > > ...remove this if: > > > >> + if (fd >= 0) > >> + close(fd); > > > > Or, do you consider it more stable in the orignal way? > > Well, as said above, this is copy/paste code. > But, anyway, when I write functions with goto statements, I prefer > having a minimal number of return statements, matter of taste. > Another way is to add two label error_close: error: but this is a bit > overkill here. All of them are OK. As for me, I prefer to not hide simple returns. > > > >> + return -1; > >> +} > >> + > >> /* Read PCI config space. */ > >> int rte_eal_pci_read_config(const struct rte_pci_device *dev, > >> void *buf, size_t len, off_t offset) > >> diff --git a/lib/librte_eal/common/eal_private.h b/lib/librte_eal/common/eal_private.h > >> index 072e672..ed1903f 100644 > >> --- a/lib/librte_eal/common/eal_private.h > >> +++ b/lib/librte_eal/common/eal_private.h > >> @@ -155,6 +155,19 @@ struct rte_pci_driver; > >> struct rte_pci_device; > >> > >> /** > >> + * Refresh a pci device object by asking the kernel for the latest information. > >> + * > >> + * This function is private to EAL. > >> + * > >> + * @param addr > >> + * The PCI Bus-Device-Function address to look for > >> + * @return > >> + * - 0 on success. > >> + * - negative on error. > > > > I don't know whether this is a convention in DPDK, anyway, I don't > > like to restrict errors to just negatives. You cannot write > > > > if ((err = pci_refresh_device(...)) /* < 0 */) { > > handle_error(err); > > } > > > > as the check for < 0 is required (easy to be avoided). > > It is a remnant of a lot of code in eal that tries to have 0 for > success, < 0 for errors, > 0 for special cases. > OK, makes sense. > -- Jan Viktorin E-mail: Viktorin@RehiveTech.com System Architect Web: www.RehiveTech.com RehiveTech Brno, Czech Republic