From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Ananyev, Konstantin" Subject: Re: [PATCH V20 1/4] bus/pci: introduce device hot unplug handle Date: Fri, 20 Apr 2018 10:32:01 +0000 Message-ID: <2601191342CEEE43887BDE71AB977258AE918A3B@IRSMSX102.ger.corp.intel.com> References: <1498711073-42917-1-git-send-email-jia.guo@intel.com> <1524058689-4954-1-git-send-email-jia.guo@intel.com> <1524058689-4954-2-git-send-email-jia.guo@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable Cc: "jblunck@infradead.org" , "shreyansh.jain@nxp.com" , "dev@dpdk.org" , "Zhang, Helin" To: "Guo, Jia" , "stephen@networkplumber.org" , "Richardson, Bruce" , "Yigit, Ferruh" , "gaetan.rivet@6wind.com" , "Wu, Jingjing" , "thomas@monjalon.net" , "motih@mellanox.com" , "matan@mellanox.com" , "Van Haaren, Harry" , "Tan, Jianfeng" Return-path: Received: from mga03.intel.com (mga03.intel.com [134.134.136.65]) by dpdk.org (Postfix) with ESMTP id 9E95C7EC7 for ; Fri, 20 Apr 2018 12:32:05 +0200 (CEST) In-Reply-To: <1524058689-4954-2-git-send-email-jia.guo@intel.com> Content-Language: en-US List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Hi Jeff, > As of device hot unplug, we need some preparatory measures so that we wil= l > not encounter memory fault after device be plug out of the system, > and also let we could recover the running data path but not been break. > This patch allows the buses to handle device hot unplug event. > The patch only enable the ops in pci bus, when handle device hot unplug > event, remap a dummy memory to avoid bus read/write error. > Other buses could accordingly implement this ops specific by themselves. >=20 > Signed-off-by: Jeff Guo > --- > v20->19: > clean the code > --- > drivers/bus/pci/pci_common.c | 67 +++++++++++++++++++++++++++= ++++++ > drivers/bus/pci/pci_common_uio.c | 32 ++++++++++++++++ > drivers/bus/pci/private.h | 12 ++++++ > lib/librte_eal/common/include/rte_bus.h | 16 ++++++++ > 4 files changed, 127 insertions(+) >=20 > diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c > index 2a00f36..709eaf3 100644 > --- a/drivers/bus/pci/pci_common.c > +++ b/drivers/bus/pci/pci_common.c > @@ -474,6 +474,72 @@ pci_find_device(const struct rte_device *start, rte_= dev_cmp_t cmp, > } >=20 > static int > +pci_handle_hot_unplug(struct rte_device *dev, void *failure_addr) > +{ > + struct rte_pci_device *pdev =3D NULL; > + int ret =3D 0, i, isfound =3D 0; > + > + if (failure_addr !=3D NULL) { > + FOREACH_DEVICE_ON_PCIBUS(pdev) { > + for (i =3D 0; i !=3D sizeof(pdev->mem_resource) / > + sizeof(pdev->mem_resource[0]); i++) { You can do i !=3D RTE_DIM(pdev->mem_resource) here. > + if ((uint64_t)failure_addr >=3D > + (uint64_t)pdev->mem_resource[i].addr && > + (uint64_t)failure_addr <=3D > + (uint64_t)pdev->mem_resource[i].addr + > + pdev->mem_resource[i].len) { I think it should be failure_addr < addr + len > + RTE_LOG(ERR, EAL, "Failure address " > + "%16.16"PRIx64" is belong to " > + "resource of device %s!\n", > + (uint64_t)failure_addr, > + pdev->device.name); > + isfound =3D 1; > + break; > + } > + } > + if (isfound) > + break; Might be it is a good thing to put the code that searches for address into = a separate function.=20 > + } > + } else if (dev !=3D NULL) { > + pdev =3D RTE_DEV_TO_PCI(dev); > + } else { > + return -EINVAL; > + } > + > + if (!pdev) > + return -1; > + > + /* remap resources for devices */ > + switch (pdev->kdrv) { > + case RTE_KDRV_VFIO: > +#ifdef VFIO_PRESENT > + /* TODO */ > +#endif Should set ret =3D-1 as not implemented now. > + break; > + case RTE_KDRV_IGB_UIO: > + case RTE_KDRV_UIO_GENERIC: > + if (rte_eal_using_phys_addrs()) { > + /* map resources for devices that use uio */ > + ret =3D pci_uio_remap_resource(pdev); > + } > + break; > + case RTE_KDRV_NIC_UIO: > + ret =3D pci_uio_remap_resource(pdev); > + break; > + default: > + RTE_LOG(DEBUG, EAL, > + " Not managed by a supported kernel driver, skipped\n"); > + ret =3D -1; > + break; > + } > + > + if (ret !=3D 0) > + RTE_LOG(ERR, EAL, "failed to handle hot unplug of %s", > + pdev->name); > + return ret; > +} > + > +static int > pci_plug(struct rte_device *dev) > { > return pci_probe_all_drivers(RTE_DEV_TO_PCI(dev)); > @@ -503,6 +569,7 @@ struct rte_pci_bus rte_pci_bus =3D { > .unplug =3D pci_unplug, > .parse =3D pci_parse, > .get_iommu_class =3D rte_pci_get_iommu_class, > + .handle_hot_unplug =3D pci_handle_hot_unplug, > }, > .device_list =3D TAILQ_HEAD_INITIALIZER(rte_pci_bus.device_list), > .driver_list =3D TAILQ_HEAD_INITIALIZER(rte_pci_bus.driver_list), > diff --git a/drivers/bus/pci/pci_common_uio.c b/drivers/bus/pci/pci_commo= n_uio.c > index 54bc20b..ba2c458 100644 > --- a/drivers/bus/pci/pci_common_uio.c > +++ b/drivers/bus/pci/pci_common_uio.c > @@ -146,6 +146,38 @@ pci_uio_unmap(struct mapped_pci_resource *uio_res) > } > } >=20 > +/* remap the PCI resource of a PCI device in anonymous virtual memory */ > +int > +pci_uio_remap_resource(struct rte_pci_device *dev) > +{ > + int i; > + void *map_address; > + > + if (dev =3D=3D NULL) > + return -1; > + > + /* Remap all BARs */ > + for (i =3D 0; i !=3D PCI_MAX_RESOURCE; i++) { > + /* skip empty BAR */ > + if (dev->mem_resource[i].phys_addr =3D=3D 0) > + continue; > + pci_unmap_resource(dev->mem_resource[i].addr, > + (size_t)dev->mem_resource[i].len); > + map_address =3D pci_map_resource( > + dev->mem_resource[i].addr, -1, 0, > + (size_t)dev->mem_resource[i].len, > + MAP_ANONYMOUS | MAP_FIXED); Instead of using mumap/mmap() can we use mremap() here? Might be a bit safer approach. > + if (map_address =3D=3D MAP_FAILED) { > + RTE_LOG(ERR, EAL, > + "Cannot remap resource for device %s\n", > + dev->name); > + return -1; > + } > + } > + > + return 0; > +} > + > static struct mapped_pci_resource * > pci_uio_find_resource(struct rte_pci_device *dev) > { > diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h > index 88fa587..cc1668c 100644 > --- a/drivers/bus/pci/private.h > +++ b/drivers/bus/pci/private.h > @@ -173,6 +173,18 @@ void pci_uio_free_resource(struct rte_pci_device *de= v, > struct mapped_pci_resource *uio_res); >=20 > /** > + * remap the pci uio resource. > + * > + * @param dev > + * Point to the struct rte pci device. > + * @return > + * - On success, zero. > + * - On failure, a negative value. > + */ > +int > +pci_uio_remap_resource(struct rte_pci_device *dev); > + > +/** > * Map device memory to uio resource > * > * This function is private to EAL. > diff --git a/lib/librte_eal/common/include/rte_bus.h b/lib/librte_eal/com= mon/include/rte_bus.h > index 6fb0834..d2c5778 100644 > --- a/lib/librte_eal/common/include/rte_bus.h > +++ b/lib/librte_eal/common/include/rte_bus.h > @@ -168,6 +168,20 @@ typedef int (*rte_bus_unplug_t)(struct rte_device *d= ev); > typedef int (*rte_bus_parse_t)(const char *name, void *addr); >=20 > /** > + * Implementation specific hot unplug handler function which is responsi= ble > + * for handle the failure when hot unplug the device, guaranty the syste= m > + * would not crash in the case. > + * @param dev > + * Pointer of the device structure. > + * > + * @return > + * 0 on success. > + * !0 on error. > + */ > +typedef int (*rte_bus_handle_hot_unplug_t)(struct rte_device *dev, > + void *dev_addr); > + > +/** > * Bus scan policies > */ > enum rte_bus_scan_mode { > @@ -209,6 +223,8 @@ struct rte_bus { > rte_bus_plug_t plug; /**< Probe single device for drivers */ > rte_bus_unplug_t unplug; /**< Remove single device from driver */ > rte_bus_parse_t parse; /**< Parse a device name */ > + rte_bus_handle_hot_unplug_t handle_hot_unplug; /**< handle hot unplug > + device event */ > struct rte_bus_conf conf; /**< Bus configuration */ > rte_bus_get_iommu_class_t get_iommu_class; /**< Get iommu class */ > }; > -- > 2.7.4