Content-Type: multipart/mixed; boundary="===============0242404790472939294==" MIME-Version: 1.0 From: Paolo Abeni To: mptcp at lists.01.org Subject: [MPTCP] Re: [PATCH 0/1] proposed export branch rebase Date: Wed, 11 Dec 2019 18:15:52 +0100 Message-ID: <321b1a9a3dc4c553f985b5b4b9d61ba685fdf779.camel@redhat.com> In-Reply-To: b4d68d34218e40cefaf811a81514ef7afae17b84.camel@redhat.com X-Status: X-Keywords: X-UID: 2872 --===============0242404790472939294== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable On Wed, 2019-12-11 at 17:32 +0100, Paolo Abeni wrote: > On Tue, 2019-12-10 at 22:27 +0100, Matthieu Baerts wrote: > > Hi, > > = > > On 10/12/2019 19:24, Matthieu Baerts wrote: > > > Hi Florian, Paolo, > > > = > > > On 09/12/2019 13:33, Paolo Abeni wrote: > > > > Hi, > > > > = > > > > On Sun, 2019-12-08 at 19:24 +0100, Florian Westphal wrote: > > > > > It includes 48 patches (down from 66), I will refrain from patch-= bombing > > > > > the list with it. There is one patch as followup here because its > > > > > "almost new" in the sense that its an amalgamation of different p= atches > > > > > that I placed in that commit because the others seemed quite larg= e = > > > > > already. > > > > > = > > > > > The __tcp_poll patch is gone. Later patches, in particular the = > > > > > changes to > > > > > only store 'tcp_sock' rather than 'struct socket' on the conn_lis= t, have > > > > > been brought forward and squashed into the relevant commits, woul= d thus > > > > > be implicitly part of basic mptcp submission. I've amemded commit > > > > > messages where appropriate. > > > > = > > > > Thank you Florian for the great work! > > > > = > > > > I went through the patches up to v1 support, and LGTM. > > > = > > > It is also good to me! Thank you very much for the very big work! The = > > > kind of work where even you you have to sit down to do it! ... O:-D > > > = > > > I just re-created the tree using your branch. I am also checking now = > > > that all commits are OK, with and without MPTCP. Also the usual = > > > compilation tests with i386 and without IPv6. > > = > > FYI, all tests are green! > > I didn't re-enable the automatic sync with net-next. I will do that = > > after having applied recent Paolo's changes. > = > The rebased and amost squashed thing is available here: > = > https://github.com/pabeni/mptcp/tree/fwestpha-mptcp_export_rebase_19_fall= back_cleanup_2 > = > It build on each commit. Not tested yet, sent-out early for review's > sake. Just tested successfully on top of V1 patches and on branch head. For the record, I observed once on the unsquashed patches, the following warn (with decoded stack trace): [ 1093.576154] ------------[ cut here ]------------ [ 1093.577188] refcount_t: addition on 0; use-after-free. [ 1093.577966] WARNING: CPU: 2 PID: 29790 at lib/refcount.c:25 refcount_war= n_saturate (/home/pabeni/devel/linux-mptcp/lib/refcount.c:25 (discriminator= 3)) [ 1093.579153] Modules linked in: sch_netem veth intel_rapl_msr rfkill inte= l_rapl_common kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_c= lmulni_intel joydev pcspkr virtio_balloon i2c_piix4 ip_tables xfs libcrc32c= crc32c_intel serio_raw virtio_net net_failover virtio_blk ata_generic fail= over virtio_console ata_piix libata [ 1093.584101] CPU: 2 PID: 29790 Comm: mptcp_connect Not tainted 5.4.0.mptc= p_bfaef67f2a47+ #464 [ 1093.585723] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS = 1.12.0-2.fc30 04/01/2014 [ 1093.587415] RIP: 0010:refcount_warn_saturate (/home/pabeni/devel/linux-m= ptcp/lib/refcount.c:25 (discriminator 3)) [ 1093.588409] Code: b9 2d 41 02 01 e8 3b 90 41 ff 0f 0b eb 9d 80 3d a8 2d = 41 02 00 75 94 48 c7 c7 c0 09 e8 87 c6 05 98 2d 41 02 01 e8 1b 90 41 ff <0f= > 0b e9 7a ff ff ff 80 3d 82 2d 41 02 00 0f 85 6d ff ff ff 48 c7 All code =3D=3D=3D=3D=3D=3D=3D=3D 0: b9 2d 41 02 01 mov $0x102412d,%ecx 5: e8 3b 90 41 ff callq 0xffffffffff419045 a: 0f 0b ud2 c: eb 9d jmp 0xffffffffffffffab e: 80 3d a8 2d 41 02 00 cmpb $0x0,0x2412da8(%rip) # 0x2412= dbd 15: 75 94 jne 0xffffffffffffffab 17: 48 c7 c7 c0 09 e8 87 mov $0xffffffff87e809c0,%rdi 1e: c6 05 98 2d 41 02 01 movb $0x1,0x2412d98(%rip) # 0x2412= dbd 25: e8 1b 90 41 ff callq 0xffffffffff419045 2a:* 0f 0b ud2 <-- trapping instruction 2c: e9 7a ff ff ff jmpq 0xffffffffffffffab 31: 80 3d 82 2d 41 02 00 cmpb $0x0,0x2412d82(%rip) # 0x2412= dba 38: 0f 85 6d ff ff ff jne 0xffffffffffffffab 3e: 48 rex.W 3f: c7 .byte 0xc7 Code starting with the faulting instruction =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D 0: 0f 0b ud2 2: e9 7a ff ff ff jmpq 0xffffffffffffff81 7: 80 3d 82 2d 41 02 00 cmpb $0x0,0x2412d82(%rip) # 0x2412= d90 e: 0f 85 6d ff ff ff jne 0xffffffffffffff81 14: 48 rex.W 15: c7 .byte 0xc7 [ 1093.592029] RSP: 0018:ffff8881061f7cd0 EFLAGS: 00010282 [ 1093.593155] RAX: 0000000000000000 RBX: ffff888035b54040 RCX: 00000000000= 00000 [ 1093.594476] RDX: 0000000000000001 RSI: 0000000000000008 RDI: ffffed1020c= 3ef8c [ 1093.595769] RBP: 0000000000000002 R08: 0000000000000001 R09: ffffed10218= 7c4d1 [ 1093.597063] R10: ffffed102187c4d0 R11: ffff88810c3e2687 R12: ffff888035b= 53fe8 [ 1093.598354] R13: ffff888035b54040 R14: ffff888035b53fc0 R15: ffff888035b= 53fc0 [ 1093.599654] FS: 00007f9b8b9ad740(0000) GS:ffff88810c200000(0000) knlGS:= 0000000000000000 [ 1093.601104] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 1093.602174] CR2: 00007fffdcdc0000 CR3: 0000000103e04001 CR4: 00000000001= 60ee0 [ 1093.603472] Call Trace: [ 1093.603997] tcp_close (/home/pabeni/devel/linux-mptcp/./include/linux/re= fcount.h:191 /home/pabeni/devel/linux-mptcp/./include/linux/refcount.h:228 = /home/pabeni/devel/linux-mptcp/./include/net/sock.h:647 /home/pabeni/devel/= linux-mptcp/net/ipv4/tcp.c:2434) [ 1093.605757] mptcp_close (/home/pabeni/devel/linux-mptcp/net/mptcp/protoc= ol.c:521 /home/pabeni/devel/linux-mptcp/net/mptcp/protocol.c:579) [ 1093.606604] inet_release (/home/pabeni/devel/linux-mptcp/net/ipv4/af_ine= t.c:428) [ 1093.607427] __sock_release (/home/pabeni/devel/linux-mptcp/net/socket.c:= 593) [ 1093.608230] sock_close (/home/pabeni/devel/linux-mptcp/net/socket.c:1272) [ 1093.608792] __fput (/home/pabeni/devel/linux-mptcp/fs/file_table.c:280) [ 1093.609430] task_work_run (/home/pabeni/devel/linux-mptcp/kernel/task_wo= rk.c:115 (discriminator 1)) [ 1093.610100] exit_to_usermode_loop (/home/pabeni/devel/linux-mptcp/./incl= ude/linux/tracehook.h:188 /home/pabeni/devel/linux-mptcp/arch/x86/entry/com= mon.c:164) [ 1093.610858] do_syscall_64 (/home/pabeni/devel/linux-mptcp/arch/x86/entry= /common.c:195 /home/pabeni/devel/linux-mptcp/arch/x86/entry/common.c:278 /h= ome/pabeni/devel/linux-mptcp/arch/x86/entry/common.c:304) [ 1093.611579] entry_SYSCALL_64_after_hwframe (/home/pabeni/devel/linux-mpt= cp/arch/x86/entry/entry_64.S:177) [ 1093.612617] RIP: 0033:0x7f9b8b4b77a0 [ 1093.613289] Code: 00 64 c7 00 0d 00 00 00 b8 ff ff ff ff eb 90 b8 ff ff = ff ff eb 89 0f 1f 40 00 83 3d fd c7 2d 00 00 75 10 b8 03 00 00 00 0f 05 <48= > 3d 01 f0 ff ff 73 31 c3 48 83 ec 08 e8 9e c5 01 00 48 89 04 24 All code =3D=3D=3D=3D=3D=3D=3D=3D 0: 00 64 c7 00 add %ah,0x0(%rdi,%rax,8) 4: 0d 00 00 00 b8 or $0xb8000000,%eax 9: ff (bad) a: ff (bad) b: ff (bad) c: ff (bad) d: eb 90 jmp 0xffffffffffffff9f f: b8 ff ff ff ff mov $0xffffffff,%eax 14: eb 89 jmp 0xffffffffffffff9f 16: 0f 1f 40 00 nopl 0x0(%rax) 1a: 83 3d fd c7 2d 00 00 cmpl $0x0,0x2dc7fd(%rip) # 0x2dc81e 21: 75 10 jne 0x33 23: b8 03 00 00 00 mov $0x3,%eax 28: 0f 05 syscall 2a:* 48 3d 01 f0 ff ff cmp $0xfffffffffffff001,%rax <--= trapping instruction 30: 73 31 jae 0x63 32: c3 retq --- But I could not reproduce it (so far) on the squashed branch. I'll suggest to merge the patches as-is and ev. look after bug after that. Cheers, Paolo --===============0242404790472939294==--