From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 0D93FA0527; Wed, 25 Nov 2020 11:01:47 +0100 (CET) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id E9054C952; Wed, 25 Nov 2020 11:01:43 +0100 (CET) Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [216.205.24.124]) by dpdk.org (Postfix) with ESMTP id 5068BC93C for ; Wed, 25 Nov 2020 11:01:41 +0100 (CET) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1606298499; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version:content-type:content-type: in-reply-to:in-reply-to:references:references; bh=vkmzNo2kc7IxQRhpQBs7U6HLKt6FBbIA+1JSsR9j0og=; b=YeSCVYduuSykBctg2dfoZHqlKoI18yH7H0Khx0F6Vc+TuYrHZE/C++eeslxqyjKPnsxKnC eLOplZeJGHbn/3Afl2DM5b6bKPFe+EK+dOmpVkvaM3MWdz0G3+/E/0F3Q7kwqhTqILQgCd c4C7x2UB06LoTySee6M9kA/LKY1UHHI= Received: from mail-vk1-f198.google.com (mail-vk1-f198.google.com [209.85.221.198]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-547-gf1b6dJBNwq4b3MdifHz1w-1; Wed, 25 Nov 2020 05:01:37 -0500 X-MC-Unique: gf1b6dJBNwq4b3MdifHz1w-1 Received: by mail-vk1-f198.google.com with SMTP id p199so254321vkp.23 for ; Wed, 25 Nov 2020 02:01:36 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=vkmzNo2kc7IxQRhpQBs7U6HLKt6FBbIA+1JSsR9j0og=; b=Hlu/3FabEeHpRE1xEhJGkbAMSpFYD0XyOAQnktjurAOfITRV8nA4P4HbGPyQ8yKPV3 5PtaJYh7X3bc9VC5IjT7o4ViwEBbjTP3NMPSRFxGJbBosnH3bM7bQpbXroFZM2px+7u0 JpoRC25JPOt/N/6ae33dQkWe/nsnM18FiZkibh1e8U8Xb3zkZYe0pCi/SDJzrcOtg5nZ jLAYxFG6vJCdaghPvTfZUbh+8ZCFY3NJRoWuh/X75gEZD9S/vTIB303KuirLejBG3h7k bExPOpKZLrTpoc0T+tlZII76PdCWOzDx0WSA853ANypnGXvjy9flAAbtIEpCBpMNThl6 UwYg== X-Gm-Message-State: AOAM533th2Gx1m7YXd30ZlQArqPebj6CodSk7V22TFtN6cS/9ox48ydB aJpX7UaYdJ4W+1Kri5tczj+wYvC0Yo6eFHETbkE74hEv/Uq4WDCaUgi/leAZ+9RRZtvCSIl2I7A ne3JbmPaISLieCBQcUdU= X-Received: by 2002:ab0:6dd3:: with SMTP id r19mr954190uaf.86.1606298496118; Wed, 25 Nov 2020 02:01:36 -0800 (PST) X-Google-Smtp-Source: ABdhPJw1ufGnj8sCs4NKgkH1Ac4OVzyNjFVPHwtlWlyhWLbR6Rgh2l5k11rv/yZutGCGy40aEYkkaUlcQ8X6Bm0Mbx8= X-Received: by 2002:ab0:6dd3:: with SMTP id r19mr954166uaf.86.1606298495749; Wed, 25 Nov 2020 02:01:35 -0800 (PST) MIME-Version: 1.0 References: <4a04b06b040ac16c45addf92fb43f59b070f185a.1606229937.git.tredaelli@redhat.com> In-Reply-To: <4a04b06b040ac16c45addf92fb43f59b070f185a.1606229937.git.tredaelli@redhat.com> From: David Marchand Date: Wed, 25 Nov 2020 11:01:25 +0100 Message-ID: To: Timothy Redaelli Cc: Anatoly Burakov , Bruce Richardson , dev , dpdk stable Authentication-Results: relay.mimecast.com; auth=pass smtp.auth=CUSA124A263 smtp.mailfrom=dmarchan@redhat.com X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset="UTF-8" Subject: Re: [dpdk-dev] [dpdk-stable] [PATCH 2/2] eal: fix loading of shared libs from driver plugin directories X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 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" On Tue, Nov 24, 2020 at 4:15 PM Timothy Redaelli wrote: > > Commit 49b536fc3060 ("eal: load only shared libs from driver plugin directories") > introduced a check that any shared library must ends with .so, but it can't > work, at least, on Fedora/CentOS/RHEL since .so symlinks are not installed > when you install dpdk package, but only when you install dpdk-devel package. > > This commit adds also a check for .so.ABI_VERSION to check for shared > lib. > > See Fedora Packaging Guidelines for more informations: > https://docs.fedoraproject.org/en-US/packaging-guidelines/#_devel_packages > > Fixes: 49b536fc3060 ("eal: load only shared libs from driver plugin directories") > Cc: bruce.richardson@intel.com > Signed-off-by: Timothy Redaelli > --- > lib/librte_eal/common/eal_common_options.c | 6 ++++-- > 1 file changed, 4 insertions(+), 2 deletions(-) > > diff --git a/lib/librte_eal/common/eal_common_options.c b/lib/librte_eal/common/eal_common_options.c > index e6f77ad217..e9e2362c53 100644 > --- a/lib/librte_eal/common/eal_common_options.c > +++ b/lib/librte_eal/common/eal_common_options.c > @@ -402,8 +402,10 @@ eal_plugindir_init(const char *path) > struct stat sb; > int nlen = strnlen(dent->d_name, sizeof(dent->d_name)); > > - /* check if name ends in .so */ > - if (strcmp(&dent->d_name[nlen - 3], ".so") != 0) > + /* check if name ends in .so or .so.ABI_VERSION */ > + if (strcmp(&dent->d_name[nlen - 3], ".so") != 0 && > + strcmp(&dent->d_name[nlen - 4 - strlen(ABI_VERSION)], > + ".so."ABI_VERSION) != 0) > continue; > > snprintf(sopath, sizeof(sopath), "%s/%s", path, dent->d_name); A first doubt I had about this patch is about multiple loading of the same driver, and its effects. So I first had a look at the main branch current state (using devtools/test-null.sh) and I can see multiple loading already happens for statically linked drivers like the bond driver. $ LD_LIBRARY_PATH=build/lib gdb build/app/dpdk-testpmd -ex "b main" -ex "run -c 3 --no-huge -m 20 -d build/drivers '--log-level=*:debug' -a 0:0.0 --vdev net_null1 --vdev net_null2 -- --no-mlockall --total-num-mbufs=2048 -ia" ... (gdb) info sharedlibrary ... 0x00007ffff796d5b0 0x00007ffff797d2d6 Yes /home/dmarchan/dpdk/build/app/../drivers/librte_net_bond.so.21 ... (gdb) b rte_eal_init (gdb) c (gdb) finish ... EAL: open shared lib build/drivers/librte_net_avp.so EAL: open shared lib build/drivers/librte_net_bond.so EAL: open shared lib build/drivers/librte_net_ixgbe.so ... (gdb) set $elm=*(struct shared_driver *)solib_list.tqh_first (gdb) while $elm p $elm set $elm=*(struct shared_driver *)($elm.next.tqe_next) end ... $67 = {next = {tqe_next = 0x5c3350, tqe_prev = 0x5c1310}, name = "build/drivers/librte_net_avp.so", '\000' , lib_handle = 0x616870} $68 = {next = {tqe_next = 0x5c4370, tqe_prev = 0x5c2330}, name = "build/drivers/librte_net_bond.so", '\000' , lib_handle = 0x7ffff7995a80} $69 = {next = {tqe_next = 0x5c5390, tqe_prev = 0x5c3350}, name = "build/drivers/librte_net_ixgbe.so", '\000' , lib_handle = 0x7ffff7942fc0} ... I could not confirm the 0x7ffff7995a80 handle points at the first load of the bond driver. gdb does not seem to know of the linker internal structures, a bit hard to tell but the addresses are in the same range, so likely the same object. I confirmed the bond constructor only being called once when the bond driver is loaded because of static link: $ LD_LIBRARY_PATH=build/lib gdb build/app/dpdk-testpmd -ex "b vdrvinitfn_pmd_bond_drv" -ex "run -c 3 --no-huge -m 20 -d build/drivers '--log-level=*:debug' -a 0:0.0 --vdev net_null1 --vdev net_null2 -- --no-mlockall --total-num-mbufs=2048 -ia" GNU gdb (GDB) Fedora 8.3.50.20190824-30.fc31 ... Reading symbols from build/app/dpdk-testpmd... Function "vdrvinitfn_pmd_bond_drv" not defined. Make breakpoint pending on future shared library load? (y or [n]) y Breakpoint 1 (vdrvinitfn_pmd_bond_drv) pending. Starting program: /home/dmarchan/dpdk/build/app/dpdk-testpmd -c 3 --no-huge -m 20 -d build/drivers '--log-level=*:debug' -a 0:0.0 --vdev net_null1 --vdev net_null2 -- --no-mlockall --total-num-mbufs=2048 -ia [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib64/libthread_db.so.1". Breakpoint 1, vdrvinitfn_pmd_bond_drv () at ../drivers/net/bonding/rte_eth_bond_pmd.c:3755 3755 RTE_PMD_REGISTER_VDEV(net_bonding, pmd_bond_drv); Missing separate debuginfos, use: dnf debuginfo-install elfutils-libelf-0.179-2.fc31.x86_64 glibc-2.30-13.fc31.x86_64 jansson-2.12-4.fc31.x86_64 libbsd-0.9.1-4.fc31.x86_64 libfdt-1.6.0-1.fc31.x86_64 numactl-libs-2.0.12-3.fc31.x86_64 zlib-1.2.11-20.fc31.x86_64 (gdb) c Continuing. ... EAL: open shared lib build/drivers/librte_net_avp.so EAL: open shared lib build/drivers/librte_net_bond.so EAL: open shared lib build/drivers/librte_net_ixgbe.so ... testpmd> Now, about this patch. In this test, it has the effect of loading all drivers twice (or thrice if statically linked). I see no problem with it since the linker is intelligent enough and constructors are only called once. My only remaining doubt is whether we should be looking for .ABI_VERSION or .ABI_MAJOR extension. This could be discussed, but in its current form, this patch lgtm. Acked-by: David Marchand -- David Marchand