From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga01.intel.com (mga01.intel.com [192.55.52.88]) by dpdk.org (Postfix) with ESMTP id 115A0903 for ; Wed, 11 Feb 2015 07:33:43 +0100 (CET) Received: from fmsmga003.fm.intel.com ([10.253.24.29]) by fmsmga101.fm.intel.com with ESMTP; 10 Feb 2015 22:33:42 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.09,556,1418112000"; d="scan'208";a="452968201" Received: from pgsmsx103.gar.corp.intel.com ([10.221.44.82]) by FMSMGA003.fm.intel.com with ESMTP; 10 Feb 2015 22:19:04 -0800 Received: from shsmsx102.ccr.corp.intel.com (10.239.4.154) by PGSMSX103.gar.corp.intel.com (10.221.44.82) with Microsoft SMTP Server (TLS) id 14.3.195.1; Wed, 11 Feb 2015 14:29:32 +0800 Received: from shsmsx101.ccr.corp.intel.com ([169.254.1.192]) by shsmsx102.ccr.corp.intel.com ([169.254.2.62]) with mapi id 14.03.0195.001; Wed, 11 Feb 2015 14:29:07 +0800 From: "Qiu, Michael" To: Tetsuya Mukawa , "Iremonger, Bernard" , "dev@dpdk.org" Thread-Topic: [dpdk-dev] [PATCH v7 04/14] eal/pci: Consolidate pci address comparison APIs Thread-Index: AQHQRELTSxWi0iDBGE+tIPrk61ioFA== Date: Wed, 11 Feb 2015 06:29:07 +0000 Message-ID: <533710CFB86FA344BFBF2D6802E60286CE7E63@SHSMSX101.ccr.corp.intel.com> References: <1422763322-13742-4-git-send-email-mukawa@igel.co.jp> <1423470639-15744-1-git-send-email-mukawa@igel.co.jp> <1423470639-15744-5-git-send-email-mukawa@igel.co.jp> <533710CFB86FA344BFBF2D6802E60286CE71EA@SHSMSX101.ccr.corp.intel.com> <8CEF83825BEC744B83065625E567D7C2049DF5CA@IRSMSX108.ger.corp.intel.com> <533710CFB86FA344BFBF2D6802E60286CE7D25@SHSMSX101.ccr.corp.intel.com> <54DAE045.6000208@igel.co.jp> <54DAE142.6090204@igel.co.jp> Accept-Language: en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [10.239.127.40] Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Subject: Re: [dpdk-dev] [PATCH v7 04/14] eal/pci: Consolidate pci address comparison APIs 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, 11 Feb 2015 06:33:44 -0000 On 2/11/2015 12:57 PM, Tetsuya Mukawa wrote:=0A= > On 2015/02/11 13:53, Tetsuya Mukawa wrote:=0A= >> On 2015/02/11 12:27, Qiu, Michael wrote:=0A= >>> On 2/10/2015 11:11 PM, Iremonger, Bernard wrote:=0A= >>>>> -----Original Message-----=0A= >>>>> From: Qiu, Michael=0A= >>>>> Sent: Monday, February 9, 2015 1:10 PM=0A= >>>>> To: Tetsuya Mukawa; dev@dpdk.org=0A= >>>>> Cc: Iremonger, Bernard=0A= >>>>> Subject: Re: [PATCH v7 04/14] eal/pci: Consolidate pci address compar= ison APIs=0A= >>>>>=0A= >>>>> On 2/9/2015 4:31 PM, Tetsuya Mukawa wrote:=0A= >>>>>> This patch replaces pci_addr_comparison() and memcmp() of pci=0A= >>>>>> addresses by eal_compare_pci_addr().=0A= >>>>>>=0A= >>>>>> v5:=0A= >>>>>> - Fix pci_scan_one to handle pt_driver correctly.=0A= >>>>>> v4:=0A= >>>>>> - Fix calculation method of eal_compare_pci_addr().=0A= >>>>>> - Add parameter checking.=0A= >>>>>>=0A= >>>>>> Signed-off-by: Tetsuya Mukawa =0A= >>>>>> ---=0A= >>>>>> lib/librte_eal/bsdapp/eal/eal_pci.c | 25 ++++++++------------= ---=0A= >>>>>> lib/librte_eal/common/eal_common_pci.c | 2 +-=0A= >>>>>> lib/librte_eal/common/include/rte_pci.h | 34 ++++++++++++++++++++= +++++++++++=0A= >>>>>> lib/librte_eal/linuxapp/eal/eal_pci.c | 25 ++++++++------------= ---=0A= >>>>>> lib/librte_eal/linuxapp/eal/eal_pci_uio.c | 2 +-=0A= >>>>>> 5 files changed, 54 insertions(+), 34 deletions(-)=0A= >>>>>>=0A= >>>>>> diff --git a/lib/librte_eal/bsdapp/eal/eal_pci.c=0A= >>>>>> b/lib/librte_eal/bsdapp/eal/eal_pci.c=0A= >>>>>> index 74ecce7..c844d58 100644=0A= >>>>>> --- a/lib/librte_eal/bsdapp/eal/eal_pci.c=0A= >>>>>> +++ b/lib/librte_eal/bsdapp/eal/eal_pci.c=0A= >>>>>> @@ -270,20 +270,6 @@ pci_uio_map_resource(struct rte_pci_device *dev= )=0A= >>>>>> return (0);=0A= >>>>>> }=0A= >>>>>>=0A= >>>>>> -/* Compare two PCI device addresses. */ -static int=0A= >>>>>> -pci_addr_comparison(struct rte_pci_addr *addr, struct rte_pci_addr= =0A= >>>>>> *addr2) -{=0A= >>>>>> - uint64_t dev_addr =3D (addr->domain << 24) + (addr->bus << 16) + (= addr->devid << 8) + addr-=0A= >>>>>> function;=0A= >>>>>> - uint64_t dev_addr2 =3D (addr2->domain << 24) + (addr2->bus << 16) = + (addr2->devid << 8) +=0A= >>>>> addr2->function;=0A= >>>>>> -=0A= >>>>>> - if (dev_addr > dev_addr2)=0A= >>>>>> - return 1;=0A= >>>>>> - else=0A= >>>>>> - return 0;=0A= >>>>>> -}=0A= >>>>>> -=0A= >>>>>> -=0A= >>>>>> /* Scan one pci sysfs entry, and fill the devices list from it. */= =0A= >>>>>> static int pci_scan_one(int dev_pci_fd, struct pci_conf *conf) @@= =0A= >>>>>> -356,13 +342,20 @@ pci_scan_one(int dev_pci_fd, struct pci_conf *con= f)=0A= >>>>>> }=0A= >>>>>> else {=0A= >>>>>> struct rte_pci_device *dev2 =3D NULL;=0A= >>>>>> + int ret;=0A= >>>>>>=0A= >>>>>> TAILQ_FOREACH(dev2, &pci_device_list, next) {=0A= >>>>>> - if (pci_addr_comparison(&dev->addr, &dev2->addr))=0A= >>>>>> + ret =3D eal_compare_pci_addr(&dev->addr, &dev2->addr);=0A= >>>>>> + if (ret > 0)=0A= >>>>>> continue;=0A= >>>>>> - else {=0A= >>>>>> + else if (ret < 0) {=0A= >>>>>> TAILQ_INSERT_BEFORE(dev2, dev, next);=0A= >>>>>> return 0;=0A= >>>>>> + } else { /* already registered */=0A= >>>>>> + /* update pt_driver */=0A= >>>>>> + dev2->pt_driver =3D dev->pt_driver;=0A= >>>>>> + free(dev);=0A= >>>>>> + return 0;=0A= >>>>>> }=0A= >>>>>> }=0A= >>>>>> TAILQ_INSERT_TAIL(&pci_device_list, dev, next); diff --git=0A= >>>>>> a/lib/librte_eal/common/eal_common_pci.c=0A= >>>>>> b/lib/librte_eal/common/eal_common_pci.c=0A= >>>>>> index f3c7f71..a89f5c3 100644=0A= >>>>>> --- a/lib/librte_eal/common/eal_common_pci.c=0A= >>>>>> +++ b/lib/librte_eal/common/eal_common_pci.c=0A= >>>>>> @@ -93,7 +93,7 @@ static struct rte_devargs *pci_devargs_lookup(stru= ct rte_pci_device *dev)=0A= >>>>>> if (devargs->type !=3D RTE_DEVTYPE_BLACKLISTED_PCI &&=0A= >>>>>> devargs->type !=3D RTE_DEVTYPE_WHITELISTED_PCI)=0A= >>>>>> continue;=0A= >>>>>> - if (!memcmp(&dev->addr, &devargs->pci.addr, sizeof(dev->addr)))= =0A= >>>>>> + if (!eal_compare_pci_addr(&dev->addr, &devargs->pci.addr))=0A= >>>>>> return devargs;=0A= >>>>>> }=0A= >>>>>> return NULL;=0A= >>>>>> diff --git a/lib/librte_eal/common/include/rte_pci.h=0A= >>>>>> b/lib/librte_eal/common/include/rte_pci.h=0A= >>>>>> index 7f2d699..4814cd7 100644=0A= >>>>>> --- a/lib/librte_eal/common/include/rte_pci.h=0A= >>>>>> +++ b/lib/librte_eal/common/include/rte_pci.h=0A= >>>>>> @@ -269,6 +269,40 @@ eal_parse_pci_DomBDF(const char *input, struct= =0A= >>>>>> rte_pci_addr *dev_addr) } #undef GET_PCIADDR_FIELD=0A= >>>>>>=0A= >>>>>> +/* Compare two PCI device addresses. */=0A= >>>>>> +/**=0A= >>>>>> + * Utility function to compare two PCI device addresses.=0A= >>>>>> + *=0A= >>>>>> + * @param addr=0A= >>>>>> + * The PCI Bus-Device-Function address to compare=0A= >>>>>> + * @param addr2=0A= >>>>>> + * The PCI Bus-Device-Function address to compare=0A= >>>>>> + * @return=0A= >>>>>> + * 0 on equal PCI address.=0A= >>>>>> + * Positive on addr is greater than addr2.=0A= >>>>>> + * Negative on addr is less than addr2, or error.=0A= >>>>>> + */=0A= >>>>>> +static inline int=0A= >>>>>> +eal_compare_pci_addr(struct rte_pci_addr *addr, struct rte_pci_addr= =0A= >>>>>> +*addr2) {=0A= >>>>>> + uint64_t dev_addr, dev_addr2;=0A= >>>>>> +=0A= >>>>>> + if ((addr =3D=3D NULL) || (addr2 =3D=3D NULL))=0A= >>>>>> + return -1;=0A= >>>>>> +=0A= >>>>>> + dev_addr =3D (addr->domain << 24) | (addr->bus << 16) |=0A= >>>>>> + (addr->devid << 8) | addr->function;=0A= >>>>>> + dev_addr2 =3D (addr2->domain << 24) | (addr2->bus << 16) |=0A= >>>>>> + (addr2->devid << 8) | addr2->function;=0A= >>>>>> +=0A= >>>>>> + if (dev_addr > dev_addr2)=0A= >>>>>> + return 1;=0A= >>>>>> + else if (dev_addr < dev_addr2)=0A= >>>>>> + return -1;=0A= >>>>>> + else=0A= >>>>>> + return 0;=0A= >>>>>> +}=0A= >>>>>> +=0A= >>>>>> /**=0A= >>>>>> * Probe the PCI bus for registered drivers.=0A= >>>>>> *=0A= >>>>>> diff --git a/lib/librte_eal/linuxapp/eal/eal_pci.c=0A= >>>>>> b/lib/librte_eal/linuxapp/eal/eal_pci.c=0A= >>>>>> index c0ca5a5..d847102 100644=0A= >>>>>> --- a/lib/librte_eal/linuxapp/eal/eal_pci.c=0A= >>>>>> +++ b/lib/librte_eal/linuxapp/eal/eal_pci.c=0A= >>>>>> @@ -229,20 +229,6 @@ error:=0A= >>>>>> return -1;=0A= >>>>>> }=0A= >>>>>>=0A= >>>>>> -/* Compare two PCI device addresses. */ -static int=0A= >>>>>> -pci_addr_comparison(struct rte_pci_addr *addr, struct rte_pci_addr= =0A= >>>>>> *addr2) -{=0A= >>>>>> - uint64_t dev_addr =3D (addr->domain << 24) + (addr->bus << 16) + (= addr->devid << 8) + addr-=0A= >>>>>> function;=0A= >>>>>> - uint64_t dev_addr2 =3D (addr2->domain << 24) + (addr2->bus << 16) = + (addr2->devid << 8) +=0A= >>>>> addr2->function;=0A= >>>>>> -=0A= >>>>>> - if (dev_addr > dev_addr2)=0A= >>>>>> - return 1;=0A= >>>>>> - else=0A= >>>>>> - return 0;=0A= >>>>>> -}=0A= >>>>>> -=0A= >>>>>> -=0A= >>>>>> /* Scan one pci sysfs entry, and fill the devices list from it. */= =0A= >>>>>> static int pci_scan_one(const char *dirname, uint16_t domain, uint8= _t=0A= >>>>>> bus, @@ -353,13 +339,20 @@ pci_scan_one(const char *dirname, uint16_= t=0A= >>>>>> domain, uint8_t bus,=0A= >>>>>> }=0A= >>>>>> else {=0A= >>>>>> struct rte_pci_device *dev2 =3D NULL;=0A= >>>>>> + int ret;=0A= >>>>>>=0A= >>>>>> TAILQ_FOREACH(dev2, &pci_device_list, next) {=0A= >>>>>> - if (pci_addr_comparison(&dev->addr, &dev2->addr))=0A= >>>>>> + ret =3D eal_compare_pci_addr(&dev->addr, &dev2->addr);=0A= >>>>>> + if (ret > 0)=0A= >>>>>> continue;=0A= >>>>>> - else {=0A= >>>>>> + else if (ret < 0) {=0A= >>>>>> TAILQ_INSERT_BEFORE(dev2, dev, next);=0A= >>>>>> return 0;=0A= >>>>>> + } else { /* already registered */=0A= >>>>>> + /* update pt_driver */=0A= >>>>>> + dev2->pt_driver =3D dev->pt_driver;=0A= >>>> Hi Tetsuya,=0A= >>>>=0A= >>>> I am seeing a problem with the librte_pmd_ixgbe code where dev->max_vf= s is being lost in some scenarios.=0A= >>>> The following line should be added here:=0A= >>>> dev2->max_vfs =3D dev->max_vfs;=0A= >>>>=0A= >>>> numa_mode should probably be updated too (although it is not causing a= problem at present).=0A= >>>> dev2->numa_mode =3D dev->numa_mode;=0A= >>> I'm very curious, why those field miss? I haven't see any places clear= =0A= >>> this field.=0A= >>>=0A= >>> What is the root cause?=0A= >> Hi Michael,=0A= >>=0A= >> Here is my guess.=0A= >> The above function creates pci device list.=0A= > I am sorry. I forgot to add below information.=0A= >=0A= > "max_vfs" or "numa_node" value is came from sysfs when the above=0A= > function is processed.=0A= =0A= Yes, but it has already been registered, why it missed?=0A= =0A= Thanks,=0A= Michael=0A= >> And current DPDK implementation assumes all devices needed to be managed= =0A= >> are under igb_uio or vfio when above code is processed.=0A= >> To add hotplug function, we also need to think some devices will start= =0A= >> to be managed under igb_uio or vfio after initializing pci device list.= =0A= >> Anyway, I guess "max_vfs" value will be changed when igb_uio or vfio=0A= >> manages the device.=0A= >>=0A= >> Hi Bernard,=0A= >>=0A= >> Could you please check "max_vfs" and "num_node" values, then check the= =0A= >> values again after the device is managed by igb_uio or vfio?=0A= >> In my environment, it seems max_vfs is created by igb_uio.=0A= >> But my NIC doesn't have VF, so behavior might be different in your=0A= >> environment.=0A= >> I guess "numa_node" should not be changed theoretically.=0A= >>=0A= >> If my guess is correct, how about replacing following values?=0A= >> - driver=0A= >> - max_vfs=0A= >> - resource=0A= >> - (numa_node)=0A= >> Except for above value, I guess other value shouldn't be changed even=0A= >> after the device is managed by igb_uio or vfio.=0A= >>=0A= >> Thanks,=0A= >> Tetsuya=0A= >>=0A= >>> Thanks,=0A= >>> Michael=0A= >>>=0A= >>>> Regards,=0A= >>>>=0A= >>>> Bernard.=0A= >>>>=0A= >>>>=0A= >>>>=0A= >>>>=0A= >>>>>> + free(dev);=0A= >>>>>> + return 0;=0A= >>>>>> }=0A= >>>>>> }=0A= >>>>>> TAILQ_INSERT_TAIL(&pci_device_list, dev, next); diff --git=0A= >>>>>> a/lib/librte_eal/linuxapp/eal/eal_pci_uio.c=0A= >>>>>> b/lib/librte_eal/linuxapp/eal/eal_pci_uio.c=0A= >>>>>> index e53f06b..1da3507 100644=0A= >>>>>> --- a/lib/librte_eal/linuxapp/eal/eal_pci_uio.c=0A= >>>>>> +++ b/lib/librte_eal/linuxapp/eal/eal_pci_uio.c=0A= >>>>>> @@ -123,7 +123,7 @@ pci_uio_map_secondary(struct rte_pci_device *dev= )=0A= >>>>>> TAILQ_FOREACH(uio_res, pci_res_list, next) {=0A= >>>>>>=0A= >>>>>> /* skip this element if it doesn't match our PCI address */=0A= >>>>>> - if (memcmp(&uio_res->pci_addr, &dev->addr, sizeof(dev->addr)))=0A= >>>>>> + if (eal_compare_pci_addr(&uio_res->pci_addr, &dev->addr))=0A= >>>>>> continue;=0A= >>>>>>=0A= >>>>>> for (i =3D 0; i !=3D uio_res->nb_maps; i++) {=0A= >>>>> Acked-by: Michael Qiu =0A= >=0A= =0A=