[PATCH 0/3] Fix several sparc64 THP bugs.

[PATCH 0/3] Fix several sparc64 THP bugs.

[David Miller]

Meelis and Aaro, I've found and fixed several THP bugs for sparc64
over the last week or so.

I cannot %100 account for the exit_mmap() WARN_ON that you two have
been able to trigger, however I'd like you both to test the changes
nonetheless.

They are against 3.15 but they should apply cleanly all the way back
to 3.13

Thanks in advance for testing.

Re: [PATCH 0/3] Fix several sparc64 THP bugs.

[mroos]

> Meelis and Aaro, I've found and fixed several THP bugs for sparc64
> over the last week or so.
> 
> I cannot %100 account for the exit_mmap() WARN_ON that you two have
> been able to trigger, however I'd like you both to test the changes
> nonetheless.
> 
> They are against 3.15 but they should apply cleanly all the way back
> to 3.13
> 
> Thanks in advance for testing.

Tried it on Netra X1, Ultra 2 and V100 that were online (applied the 
patches and enabled THP with defaulting to always).

Ultra 2 did not boot up (will see on Monday).

Netra X1 performed a simple dist-upgrade fine with this kernel.

V100 boots up fine but as soon as I start aptitude, it just hangs with 
nothing on console (tried it twice).

-- 
Meelis Roos (mroos@linux.ee)

Re: [PATCH 0/3] Fix several sparc64 THP bugs.

[David Miller]

From: mroos@linux.ee
Date: Sat, 26 Apr 2014 23:14:59 +0300 (EEST)

>> Meelis and Aaro, I've found and fixed several THP bugs for sparc64
>> over the last week or so.
>> 
>> I cannot %100 account for the exit_mmap() WARN_ON that you two have
>> been able to trigger, however I'd like you both to test the changes
>> nonetheless.
>> 
>> They are against 3.15 but they should apply cleanly all the way back
>> to 3.13
>> 
>> Thanks in advance for testing.
> 
> Tried it on Netra X1, Ultra 2 and V100 that were online (applied the 
> patches and enabled THP with defaulting to always).
> 
> Ultra 2 did not boot up (will see on Monday).
> 
> Netra X1 performed a simple dist-upgrade fine with this kernel.
> 
> V100 boots up fine but as soon as I start aptitude, it just hangs with 
> nothing on console (tried it twice).

Sigh, thanks for testing...

Re: [PATCH 0/3] Fix several sparc64 THP bugs.

[Meelis Roos]

> >> Meelis and Aaro, I've found and fixed several THP bugs for sparc64
> >> over the last week or so.
> >> 
> >> I cannot %100 account for the exit_mmap() WARN_ON that you two have
> >> been able to trigger, however I'd like you both to test the changes
> >> nonetheless.
> >> 
> >> They are against 3.15 but they should apply cleanly all the way back
> >> to 3.13
> >> 
> >> Thanks in advance for testing.
> > 
> > Tried it on Netra X1, Ultra 2 and V100 that were online (applied the 
> > patches and enabled THP with defaulting to always).
> > 
> > Ultra 2 did not boot up (will see on Monday).

When I rebooted it, the hang is not reproducible - Debian dist-upgrade 
went well. However, it is still strange.

-- 
Meelis Roos (mroos@linux.ee)

Re: [PATCH 0/3] Fix several sparc64 THP bugs.

[Meelis Roos]

> V100 boots up fine but as soon as I start aptitude, it just hangs with 
> nothing on console (tried it twice).

Got this on V100 after retries (truncated to width by minicom):

[ 1286.980912] BUG: soft lockup - CPU#0 stuck for 22s! [aptitude:2520]
[ 1287.063316] Modules linked in: ipv6 loop fuse ohci_pci ohci_hcd usbcore sg sr_mod cdrom usb_commn
[ 1287.181103] irq event stamp: 23310
[ 1287.225692] hardirqs last  enabled at (23309): [<0000000000404b1c>] rtrap_xcall+0x18/0x20
[ 1287.333358] hardirqs last disabled at (23310): [<0000000000426b00>] valid_addr_bitmap_patch+0x388
[ 1287.455764] softirqs last  enabled at (23308): [<000000000045916c>] __do_softirq+0x18c/0x240
[ 1287.566730] softirqs last disabled at (23301): [<000000000042b9e0>] do_softirq_own_stack+0x20/0x0
[ 1287.684574] CPU: 0 PID: 2520 Comm: aptitude Not tainted 3.15.0-rc2-dirty #115
[ 1287.778385] task: ffffff006e230fc0 ti: ffffff006c428000 task.ti: ffffff006c428000
[ 1287.876805] TSTATE: 0000000080001604 TPC: 000000000040796c TNPC: 0000000000407970 Y: 00000ef4   d
[ 1288.006079] TPC: <sparc64_realfault_common+0x8/0x20>
[ 1288.071290] g0: ffffff006e0a72f0 g1: 0000000000000000 g2: 0000000000000000 g3: 0000000000000598
[ 1288.185720] g4: ffffff006e230fc0 g5: 000000000000ff00 g6: ffffff006c428000 g7: 0000000000000030
[ 1288.300122] o0: 0000000000000001 o1: ffffff006c428400 o2: 0000000000637c20 o3: 0000000000000ef4
[ 1288.414524] o4: ffffff006e559240 o5: 0000000000000003 sp: ffffff006c42b491 ret_pc: 0000000000404c
[ 1288.533500] RPC: <rtrap_xcall+0x18/0x20>
[ 1288.585002] l0: 0000000000001000 l1: 0000000011001603 l2: 000000000040796c l3: 0000000000000000
[ 1288.699432] l4: 0000000000000003 l5: 00000000ffa2ce8b l6: ffffff006c428000 l7: 0000000080001004
[ 1288.813834] i0: 00000000ffa2cf28 i1: ffffff006c428048 i2: 0000000000000000 i3: 0000010093604ee0
[ 1288.928236] i4: 00000000ffa2cf28 i5: 0000000000000000 i6: ffffff006c42b5f1 i7: 000000000042c728
[ 1289.042645] I7: <fault_in_user_windows+0x68/0x100>
[ 1289.105570] Call Trace:
[ 1289.137685]  [0000000000637c1c] ___copy_to_user+0x121c/0x13c0
[ 1289.213230]  [0000000000404a90] __handle_user_windows+0x0/0x10
[ 1289.290271] Kernel panic - not syncing: softlockup: hung tasks
[ 1289.366915] CPU: 0 PID: 2520 Comm: aptitude Not tainted 3.15.0-rc2-dirty #115
[ 1289.460732] Call Trace:
[ 1289.492836]  [00000000007e1290] panic+0xb8/0x244
[ 1289.553522]  [00000000004bcd7c] watchdog_timer_fn+0x13c/0x160
[ 1289.629074]  [000000000047930c] __run_hrtimer.isra.20+0x4c/0x140
[ 1289.708035]  [0000000000479928] hrtimer_interrupt+0xc8/0x240
[ 1289.782434]  [000000000042f730] timer_interrupt+0x50/0xa0
[ 1289.853398]  [0000000000426b0c] valid_addr_bitmap_patch+0x44/0x1b8
[ 1289.934653]  [000000000040796c] sparc64_realfault_common+0x8/0x20
[ 1290.014770]  [0000000000637c1c] ___copy_to_user+0x121c/0x13c0
[ 1290.090304]  [0000000000404a90] __handle_user_windows+0x0/0x10

-- 
Meelis Roos (mroos@ut.ee)      http://www.cs.ut.ee/~mroos/