Message ID | CAHp75VdhaZrzBrtHrFCvsckpGq=oLvCRtR21CQH7C8XduMoA_w@mail.gmail.com |
---|---|
State | RFC, archived |
Delegated to: | David Miller |
Headers | show |
How much memory you have on this host ? dmesg | grep hash On Fri, Jul 31, 2015 at 3:45 PM, Andy Shevchenko <andy.shevchenko@gmail.com> wrote: > On Fri, Jul 31, 2015 at 4:25 PM, Andy Shevchenko > <andy.shevchenko@gmail.com> wrote: >> Hi! >> >> Got few weeks ago an old AVR32 board (ATNGW100). >> It has ethernet cards supported by macb driver. >> >> Bring it mostly back to work with recent kernel from linux-next. Now, >> when I start networking on it, I got in few seconds kernel panic. > > Seems the hack fixes this (still playing with network connected). > > # uname -a > Linux buildroot 4.2.0-rc4-next-20150731+ #164 Fri Jul 31 16:37:20 EEST > 2015 avr32 GNU/Linux > > > --- a/net/ipv4/udp.c > +++ b/net/ipv4/udp.c > @@ -1665,6 +1665,10 @@ static int __udp4_lib_mcast_deliver(struct net > *net, struct sk_buff *skb, > unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10); > bool inner_flushed = false; > > +#ifdef CONFIG_AVR32 > + use_hash2 = 0; > +#endif > + > if (use_hash2) { > hash2_any = udp4_portaddr_hash(net, htonl(INADDR_ANY), hnum) & > udp_table.mask; > > > >> >> Unable to handle kernel NULL pointer dereference at virtual address 00000000 >> ptbr = 91e42000 pgd = 91e4b000 >> Oops: Kernel access of bad area, sig: 11 [#1] >> FRAME_POINTER chip: 0x01f:0x1e82 rev 2 >> Modules linked in: >> CPU: 0 PID: 0 Comm: swapper Not tainted 4.2.0-rc4-next-20150729+ #102 >> task: 903532dc ti: 90350000 task.ti: 90350000 >> PC is at __udp4_lib_rcv+0x300/0x660 >> LR is at 0x15da8a2f >> pc : [<90233a84>] lr : [<15da8a2f>] Not tainted >> 0sp : 90351b5c r12: 00000000 r11: 91cd4450 >> r10: 00000000 r9 : 0000004c r8 : 00000000 >> r7 : 90351c80 r6 : 91db5c80 r5 : 91e4e540 r4 : 11f6a114 >> r3 : 00000000 r2 : 91e6b224 r1 : 0000008a r0 : 00000000 >> Flags: qvnZc >> Mode bits: hjmde....g >> CPU Mode: Interrupt level 0 >> Stack: (0x90351b5c to 0x90352000) >> … >> Call trace: >> [<90233df0>] udp_rcv+0xc/0x14 >> [<90215220>] ip_local_deliver_finish+0xac/0x15c >> [<902150ca>] ip_local_deliver+0x76/0x84 >> [<90214d32>] ip_rcv_finish+0x23a/0x250 >> [<90215004>] ip_rcv+0x2bc/0x30c >> [<901f5ee4>] __netif_receive_skb_core+0x548/0x570 >> [<901f5f52>] __netif_receive_skb+0x46/0x50 >> [<901f5f8e>] netif_receive_skb_internal+0x32/0x3c >> [<901f5fa0>] netif_receive_skb_sk+0x8/0xc >> [<901cf358>] macb_rx+0x1b0/0x1d8 >> [<901cf4d8>] macb_poll+0x38/0xa4 >> [<901faf9c>] net_rx_action+0x84/0x1b4 >> [<90021e46>] __do_softirq+0x5a/0x150 >> [<90021f86>] irq_exit+0x26/0x58 >> [<9001a520>] do_IRQ+0x34/0x44 >> [<90019428>] irq_level0+0x18/0x5c >> [<900355a4>] default_idle_call+0x1c/0x20 >> [<9003563e>] cpu_startup_entry+0x66/0xa8 >> [<902a16a4>] rest_init+0x48/0x70 >> [<900007fc>] start_kernel+0x290/0x2dc >> >> Long time bisecting and reading assembly points to the >> >> commit 2dc41cff7545d55c6294525c811594576f8e119c >> Author: David Held <drheld@google.com> >> Date: Tue Jul 15 23:28:32 2014 -0400 >> >> udp: Use hash2 for long hash1 chains in __udp*_lib_mcast_deliver. >> >> I don't know yet neither the package exactly makes this (tried to >> debug print, but bug disappears) nor should be network card driver >> fixed, or even compiler / binutils problems (using it from buildroot, >> which is gcc 4.2.2). Would like to hear opinions what the root cause >> might be and what ways we have to fix it. I'm also wondering if any >> other architecture with same network card has the issue. >> >> P.S. >> Since buildroot is not supporting anymore avr32 I use the version just >> before this removal. (Nicolas, does Atmel care about that?) >> >> -- >> With Best Regards, >> Andy Shevchenko > > > > -- > With Best Regards, > Andy Shevchenko -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, Jul 31, 2015 at 4:49 PM, Eric Dumazet <edumazet@google.com> wrote: > How much memory you have on this host ? > > dmesg | grep hash > > # dmesg | grep hash PID hash table entries: 128 (order: -3, 512 bytes) Dentry cache hash table entries: 4096 (order: 2, 16384 bytes) Inode-cache hash table entries: 2048 (order: 1, 8192 bytes) Mount-cache hash table entries: 1024 (order: 0, 4096 bytes) Mountpoint-cache hash table entries: 1024 (order: 0, 4096 bytes) TCP established hash table entries: 1024 (order: 0, 4096 bytes) TCP bind hash table entries: 1024 (order: 0, 4096 bytes) UDP hash table entries: 256 (order: 0, 4096 bytes) UDP-Lite hash table entries: 256 (order: 0, 4096 bytes) futex hash table entries: 16 (order: -5, 192 bytes) > On Fri, Jul 31, 2015 at 3:45 PM, Andy Shevchenko > <andy.shevchenko@gmail.com> wrote: >> On Fri, Jul 31, 2015 at 4:25 PM, Andy Shevchenko >> <andy.shevchenko@gmail.com> wrote: >>> Hi! >>> >>> Got few weeks ago an old AVR32 board (ATNGW100). >>> It has ethernet cards supported by macb driver. >>> >>> Bring it mostly back to work with recent kernel from linux-next. Now, >>> when I start networking on it, I got in few seconds kernel panic. >> >> Seems the hack fixes this (still playing with network connected). >> >> # uname -a >> Linux buildroot 4.2.0-rc4-next-20150731+ #164 Fri Jul 31 16:37:20 EEST >> 2015 avr32 GNU/Linux >> >> >> --- a/net/ipv4/udp.c >> +++ b/net/ipv4/udp.c >> @@ -1665,6 +1665,10 @@ static int __udp4_lib_mcast_deliver(struct net >> *net, struct sk_buff *skb, >> unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10); >> bool inner_flushed = false; >> >> +#ifdef CONFIG_AVR32 >> + use_hash2 = 0; >> +#endif >> + >> if (use_hash2) { >> hash2_any = udp4_portaddr_hash(net, htonl(INADDR_ANY), hnum) & >> udp_table.mask; >> >> >> >>> >>> Unable to handle kernel NULL pointer dereference at virtual address 00000000 >>> ptbr = 91e42000 pgd = 91e4b000 >>> Oops: Kernel access of bad area, sig: 11 [#1] >>> FRAME_POINTER chip: 0x01f:0x1e82 rev 2 >>> Modules linked in: >>> CPU: 0 PID: 0 Comm: swapper Not tainted 4.2.0-rc4-next-20150729+ #102 >>> task: 903532dc ti: 90350000 task.ti: 90350000 >>> PC is at __udp4_lib_rcv+0x300/0x660 >>> LR is at 0x15da8a2f >>> pc : [<90233a84>] lr : [<15da8a2f>] Not tainted >>> 0sp : 90351b5c r12: 00000000 r11: 91cd4450 >>> r10: 00000000 r9 : 0000004c r8 : 00000000 >>> r7 : 90351c80 r6 : 91db5c80 r5 : 91e4e540 r4 : 11f6a114 >>> r3 : 00000000 r2 : 91e6b224 r1 : 0000008a r0 : 00000000 >>> Flags: qvnZc >>> Mode bits: hjmde....g >>> CPU Mode: Interrupt level 0 >>> Stack: (0x90351b5c to 0x90352000) >>> … >>> Call trace: >>> [<90233df0>] udp_rcv+0xc/0x14 >>> [<90215220>] ip_local_deliver_finish+0xac/0x15c >>> [<902150ca>] ip_local_deliver+0x76/0x84 >>> [<90214d32>] ip_rcv_finish+0x23a/0x250 >>> [<90215004>] ip_rcv+0x2bc/0x30c >>> [<901f5ee4>] __netif_receive_skb_core+0x548/0x570 >>> [<901f5f52>] __netif_receive_skb+0x46/0x50 >>> [<901f5f8e>] netif_receive_skb_internal+0x32/0x3c >>> [<901f5fa0>] netif_receive_skb_sk+0x8/0xc >>> [<901cf358>] macb_rx+0x1b0/0x1d8 >>> [<901cf4d8>] macb_poll+0x38/0xa4 >>> [<901faf9c>] net_rx_action+0x84/0x1b4 >>> [<90021e46>] __do_softirq+0x5a/0x150 >>> [<90021f86>] irq_exit+0x26/0x58 >>> [<9001a520>] do_IRQ+0x34/0x44 >>> [<90019428>] irq_level0+0x18/0x5c >>> [<900355a4>] default_idle_call+0x1c/0x20 >>> [<9003563e>] cpu_startup_entry+0x66/0xa8 >>> [<902a16a4>] rest_init+0x48/0x70 >>> [<900007fc>] start_kernel+0x290/0x2dc >>> >>> Long time bisecting and reading assembly points to the >>> >>> commit 2dc41cff7545d55c6294525c811594576f8e119c >>> Author: David Held <drheld@google.com> >>> Date: Tue Jul 15 23:28:32 2014 -0400 >>> >>> udp: Use hash2 for long hash1 chains in __udp*_lib_mcast_deliver. >>> >>> I don't know yet neither the package exactly makes this (tried to >>> debug print, but bug disappears) nor should be network card driver >>> fixed, or even compiler / binutils problems (using it from buildroot, >>> which is gcc 4.2.2). Would like to hear opinions what the root cause >>> might be and what ways we have to fix it. I'm also wondering if any >>> other architecture with same network card has the issue. >>> >>> P.S. >>> Since buildroot is not supporting anymore avr32 I use the version just >>> before this removal. (Nicolas, does Atmel care about that?) >>> >>> -- >>> With Best Regards, >>> Andy Shevchenko >> >> >> >> -- >> With Best Regards, >> Andy Shevchenko
--- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -1665,6 +1665,10 @@ static int __udp4_lib_mcast_deliver(struct net *net, struct sk_buff *skb, unsigned int hash2 = 0, hash2_any = 0, use_hash2 = (hslot->count > 10); bool inner_flushed = false; +#ifdef CONFIG_AVR32 + use_hash2 = 0; +#endif + if (use_hash2) { hash2_any = udp4_portaddr_hash(net, htonl(INADDR_ANY), hnum) &