From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mails.dpdk.org (mails.dpdk.org [217.70.189.124]) by inbox.dpdk.org (Postfix) with ESMTP id E71DBA0C4B; Sat, 14 Aug 2021 00:10:17 +0200 (CEST) Received: from [217.70.189.124] (localhost [127.0.0.1]) by mails.dpdk.org (Postfix) with ESMTP id 68EB740DDE; Sat, 14 Aug 2021 00:10:17 +0200 (CEST) Received: from mail-il1-f178.google.com (mail-il1-f178.google.com [209.85.166.178]) by mails.dpdk.org (Postfix) with ESMTP id A23584013F for ; Sat, 14 Aug 2021 00:10:15 +0200 (CEST) Received: by mail-il1-f178.google.com with SMTP id v16so1594073ilo.10 for ; Fri, 13 Aug 2021 15:10:15 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=banduracyber-com.20150623.gappssmtp.com; s=20150623; h=subject:to:cc:references:from:message-id:date:user-agent :mime-version:in-reply-to:content-language; bh=Af1ectn0hNIatuXJh/A21t2eBF8tFwif4OrOtXNpzK8=; b=sZlJnyZZ/tJjg+X57lw8ByMB1wNP8FruKux39JEJkZ388tUrQfT6NLpV92iqSH0fPP Prcgbf/ELUeyweCzsD2T5VTvBoQ6xZeOy5H3A5l/IF0MsLSI2mBWG/pEPDJKlavrVYLK o3E8w9cphffCKJHtNiGyjmvHYzCS7KsnnjWMaf4mxS9Ji88hdHLDdtr+znM31Pa8gEzp DyHIhCPWtKko9auW4PSs9l2zJrIskldIZyWtGU1z/iAhnz10GYJp1eiJAF+8w0UT+/5i LrzaUPFVH0KIjtEz7ifMoaQQLtrdLgZy7AN9bR2kFIR2Ke8lvA3LgltPVAfF/NorPQev FNFQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:subject:to:cc:references:from:message-id:date :user-agent:mime-version:in-reply-to:content-language; bh=Af1ectn0hNIatuXJh/A21t2eBF8tFwif4OrOtXNpzK8=; b=rVQKmKM4CPOIsp5IOnRk99CxJ+7VroR1Ti7sy4n3gUgzOcM7ChuCUPO6M+5r3NBlUE 2+ZppvBUfpsXisD1vCTdxFG5qkSaDRnDANJozpwu0cpe/EzUlqzWAUlZeNZXd6NGLlyC WuV5/+OllbhIVy5iBpXah7zxui81jpJa94znVba1UHv+F95z1CyNnASIbGmmh7Fi2Qh0 JR7glnwFSqyVaaUApgxq0rNxXiwDMOi+F5PnMDV0Zn4WnQeW7uozFoe1IiuMs6pvutmq CXX3+cdRWpy/uTVx3ndVrSfWvwSiN9n/2IiN7BOhF5Wnz7CzFCl7uQMKGOfCG7WS36iN LsVw== X-Gm-Message-State: AOAM530pPJSDfGLQXBCrqhzCCKUUFQCD4c6cHkeqkYIBFNlbn3DyoWWb QuMLqLeuPBqyfTrd8xmPUfNcIw== X-Google-Smtp-Source: ABdhPJwC0mqjM1IfIq7nRqKlzIluCn8IrU7hP7GdVFbt4F2wtk43mmovlF1yejvtWCuGz0kTRsNSbg== X-Received: by 2002:a05:6e02:b4b:: with SMTP id f11mr3252769ilu.130.1628892614846; Fri, 13 Aug 2021 15:10:14 -0700 (PDT) Received: from [192.168.0.3] (dyn-69-88-36-66.myactv.net. [69.88.36.66]) by smtp.gmail.com with ESMTPSA id e2sm1628672ioc.0.2021.08.13.15.10.13 (version=TLS1_3 cipher=TLS_AES_128_GCM_SHA256 bits=128/128); Fri, 13 Aug 2021 15:10:14 -0700 (PDT) To: Long Li , Stephen Hemminger Cc: "dev@dpdk.org" , "stable@dpdk.org" References: <20210726170040.25155-1-jonathan.erb@banduracyber.com> From: Jonathan Erb Message-ID: <4387890b-e4da-9a93-a142-dfd976b291b0@banduracyber.com> Date: Fri, 13 Aug 2021 18:10:13 -0400 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.11.0 MIME-Version: 1.0 In-Reply-To: Content-Language: en-US Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 8bit X-Content-Filtered-By: Mailman/MimeDel 2.1.29 Subject: Re: [dpdk-dev] [PATCH v2] bus/vmbus: Fix crash when handling packets in secondary process X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" The first part makes sense. There are several ways to iterate over all the subchannels. I'm probably not fully understanding the second part. vmbus_uio_map_secondary_subchan() wants to map to br->vbr but that does not appear to be initialized anywhere. In fact, it looks like we should set br->vbr to mapaddr which is what vmbus_uio_map_subchan() does. Jonathan On 8/11/21 6:06 PM, Long Li wrote: > > I think the code is on the right track. > > Instead of using vmbus_uio_get_num_subchan() and calling > vmbus_uio_get_subchan() on each channel, you can just create a new > function vmbus_uio_get_secondary_subchan(). This function goes through > all subchannels and map ring buffers to the same addresses used by the > primary process. > > In the original code, vmbus_uio_map_secondary_subchan() has a check > for “if (mapaddr == ring_buf)”. If the address is mapped to somewhere > else, the address won’t work for the secondary process. So we need to > keep this check. > > *From:* Stephen Hemminger > *Sent:* Wednesday, August 11, 2021 8:26 AM > *To:* Jonathan Erb ; Long Li > > *Cc:* dev@dpdk.org; stable@dpdk.org > *Subject:* RE: [PATCH v2] bus/vmbus: Fix crash when handling packets > in secondary process > > Looks fine, only comments would be to keep to original style and add > check if primary channel not found? > > *From:* Jonathan Erb > > *Sent:* Monday, August 9, 2021 9:07 AM > *To:* Long Li >; > Stephen Hemminger > > *Cc:* dev@dpdk.org ; stable@dpdk.org > > *Subject:* Re: [PATCH v2] bus/vmbus: Fix crash when handling packets > in secondary process > > > > > You don't often get email from jonathan.erb@banduracyber.com > . Learn why this is important > > > > > Would it be possible to resolve the lack of subchannels in secondary > processes as follows: > > 1. Create the following function in vmbus_uio.c to obtain the count of > subchannels created by the primary: > > int vmbus_uio_get_num_subchan(struct vmbus_channel *primary) > { > >     const struct rte_vmbus_device *dev = primary->device; >     char chan_path[PATH_MAX]; >     struct dirent *ent; >     DIR *chan_dir; >     int err; >     int subchan_cnt = 0; > >     snprintf(chan_path, sizeof(chan_path), >          "%s/%s/channels", >          SYSFS_VMBUS_DEVICES, dev->device.name); > >     chan_dir = opendir(chan_path); >     if (!chan_dir) { >         VMBUS_LOG(ERR, "cannot open %s: %s", >               chan_path, strerror(errno)); >         return 0; >     } > >     while ((ent = readdir(chan_dir))) { >         unsigned long relid, subid, monid; >         char *endp; > >         if (ent->d_name[0] == '.') >             continue; > >         errno = 0; >         relid = strtoul(ent->d_name, &endp, 0); >         if (*endp || errno != 0 || relid > UINT16_MAX) { >             VMBUS_LOG(NOTICE, "not a valid channel relid: %s", >                   ent->d_name); >             continue; >         } >         subchan_cnt++; >     } > >     closedir(chan_dir); >     //Less one for primary channel >     return subchan_cnt - 1; > > } > > 2. Then change the bottom of vmbus_uio_map_secondary() to be simply: > >     /* fd is not needed in secondary process, close it */ >     close(fd); > >     if (vmbus_chan_create(dev, dev->relid, 0, >                     dev->monitor_id, &dev->primary)) { >         VMBUS_LOG(ERR, "cannot create primary channel"); >         return -1; >     } > >     int subchannels_cnt = vmbus_uio_get_num_subchan(dev->primary); >     for (i = 0; i < subchannels_cnt; i++) { >         if(vmbus_uio_get_subchan(dev->primary, &chan)) >             return -1; > STAILQ_INSERT_TAIL(&dev->primary->subchannel_list, chan, next); >     } >     return 0; > } > > In this way the secondary processes detect the number of subchannels > created by the primary, then perform their own mappings as needed. All > this can occur before rte_eal_init has completed. > > Jonathan > > On 7/26/21 6:16 PM, Long Li wrote: > > Subject: [PATCH v2] bus/vmbus: Fix crash when handling packets in > > secondary process > > Have secondary processes construct their own copy of primary channel with > > own mappings. > > Remove vmbus_channel primary ptr from struct mapped_vmbus_resource > > as its not used. > > Populate virtual memory address "addr" in struct rte_mem_resource for > > secondary processes as netvsc will attempt to reference it thus causing a > > crash. It was initialized for primary processes but not for secondary. > > Cc:stable@dpdk.org > > Signed-off-by: Jonathan Erb > > --- > > v2: > > * Remove unnecessary check for NULL pointer before call to rte_free() per > > reviwer comment. > > drivers/bus/vmbus/private.h          |  1 - > > drivers/bus/vmbus/vmbus_channel.c    |  4 +--- > > drivers/bus/vmbus/vmbus_common_uio.c | 14 +++++++++----- > > 3 files changed, 10 insertions(+), 9 deletions(-) > > diff --git a/drivers/bus/vmbus/private.h b/drivers/bus/vmbus/private.h > > index 528d60a42f..746212bd5f 100644 > > --- a/drivers/bus/vmbus/private.h > > +++ b/drivers/bus/vmbus/private.h > > @@ -42,7 +42,6 @@ struct mapped_vmbus_resource { > >   rte_uuid_t id; > >   int nb_maps; > > -  struct vmbus_channel *primary; > >   struct vmbus_map maps[VMBUS_MAX_RESOURCE]; > >   char path[PATH_MAX]; > > }; > > diff --git a/drivers/bus/vmbus/vmbus_channel.c > > b/drivers/bus/vmbus/vmbus_channel.c > > index f67f1c438a..119b9b367e 100644 > > --- a/drivers/bus/vmbus/vmbus_channel.c > > +++ b/drivers/bus/vmbus/vmbus_channel.c > > @@ -351,10 +351,8 @@ int rte_vmbus_chan_open(struct rte_vmbus_device > > *device, > >   err = vmbus_chan_create(device, device->relid, 0, > >                          device->monitor_id, new_chan); > > -  if (!err) { > > +  if (!err) > >           device->primary = *new_chan; > > -          uio_res->primary = *new_chan; > > -  } > >   return err; > > } > > diff --git a/drivers/bus/vmbus/vmbus_common_uio.c > > b/drivers/bus/vmbus/vmbus_common_uio.c > > index 8582e32c1d..83c56b6fa2 100644 > > --- a/drivers/bus/vmbus/vmbus_common_uio.c > > +++ b/drivers/bus/vmbus/vmbus_common_uio.c > > @@ -69,8 +69,10 @@ vmbus_uio_map_secondary(struct rte_vmbus_device > > *dev) > >                                       fd, offset, > >                                       uio_res->maps[i].size, 0); > > -          if (mapaddr == uio_res->maps[i].addr) > > +          if (mapaddr == uio_res->maps[i].addr) { > > +                  dev->resource[i].addr = mapaddr; > >                   continue;      /* successful map */ > > +          } > >           if (mapaddr == MAP_FAILED) > >                   VMBUS_LOG(ERR, > > @@ -88,9 +90,9 @@ vmbus_uio_map_secondary(struct rte_vmbus_device > > *dev) > >   /* fd is not needed in secondary process, close it */ > >   close(fd); > > -  dev->primary = uio_res->primary; > > -  if (!dev->primary) { > > -          VMBUS_LOG(ERR, "missing primary channel"); > > +  if (vmbus_chan_create(dev, dev->relid, 0, > > +                                 dev->monitor_id, &dev->primary)) { > > +          VMBUS_LOG(ERR, "cannot create primary channel"); > >           return -1; > >   } > > Looking at this closer, I don't think it will work for subchannels in the secondary process. > > The code after is: > >         STAILQ_FOREACH(chan, &dev->primary->subchannel_list, next) { > >                 if (vmbus_uio_map_secondary_subchan(dev, chan) != 0) { > >                         VMBUS_LOG(ERR, "cannot map secondary subchan"); > >                         return -1; > >                 } > >         } > > Because at this time, the primary channel is just created, "subchannel_list" should be NULL. The secondary process ends up running without subchannels. In the primary process, "subchannel_list" are populated when it calls hn_dev_configure(), so it is good. > > Sorry I didn't spot this earlier. The best way to fix this is to use rte memory functions for allocating vmbus device.  In this way they can be properly mapped to the secondary process. But rte memory functions are not available at the time vmbus device is probed. I haven't been able to find a good way to fix this. Will keep looking. > > Thanks, > > Long > > @@ -211,8 +213,10 @@ vmbus_uio_unmap_resource(struct > > rte_vmbus_device *dev) > >           return; > >   /* secondary processes - just free maps */ > > -  if (rte_eal_process_type() != RTE_PROC_PRIMARY) > > +  if (rte_eal_process_type() != RTE_PROC_PRIMARY) { > > +          rte_free(dev->primary); > >           return vmbus_uio_unmap(uio_res); > > +  } > >   TAILQ_REMOVE(uio_res_list, uio_res, next); > > -- > > 2.17.1 >