<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<br>
<br>
<div class="moz-cite-prefix">On 3/14/20 9:18 AM, Nicholas Piggin
wrote:<br>
</div>
<blockquote type="cite"
cite="mid:1584157063.g5s75uhbdu.astroid@bobo.none">
<pre class="moz-quote-pre" wrap="">Ganesh Goudar's on March 14, 2020 12:04 am:
</pre>
<blockquote type="cite">
<pre class="moz-quote-pre" wrap="">MCE handling on pSeries platform fails as recent rework to use common
code for pSeries and PowerNV in machine check error handling tries to
access per-cpu variables in realmode. The per-cpu variables may be
outside the RMO region on pSeries platform and needs translation to be
enabled for access. Just moving these per-cpu variable into RMO region
did'nt help because we queue some work to workqueues in real mode, which
again tries to touch per-cpu variables.
</pre>
</blockquote>
<pre class="moz-quote-pre" wrap="">
Which queues are these? We should not be using Linux workqueues, but the
powerpc mce code which uses irq_work.</pre>
</blockquote>
<pre>Yes, irq work queues accesses memory outside RMO.
irq_work_queue()->__irq_work_queue_local()->[this_cpu_ptr(&lazy_list) | this_cpu_ptr(&raised_list)]
</pre>
<blockquote type="cite"
cite="mid:1584157063.g5s75uhbdu.astroid@bobo.none">
<blockquote type="cite">
<pre class="moz-quote-pre" wrap="">Also fwnmi_release_errinfo()
cannot be called when translation is not enabled.
</pre>
</blockquote>
<pre class="moz-quote-pre" wrap="">
Why not?</pre>
</blockquote>
<pre class="moz-quote-pre" wrap="">It crashes when we try to get RTAS token for "ibm, nmi-interlock" device
tree node. But yes we can avoid it by storing it rtas_token somewhere but haven't
tried it, here is the backtrace I got when fwnmi_release_errinfo() called from
realmode handler.
[ 70.856908] BUG: Unable to handle kernel data access on read at 0xc0000001ffffa8f8
[ 70.856918] Faulting instruction address: 0xc000000000853920
[ 70.856927] Oops: Kernel access of bad area, sig: 11 [#1]
[ 70.856935] LE PAGE_SIZE=64K MMU=Hash SMP NR_CPUS=2048 NUMA pSeries
[ 70.856943] Modules linked in: mcetest_slb(OE+) bridge stp llc ebtable_filter ebtables ip6table_filter ip6_tables iptable_filter sg pseries_rng ip_tables xfs libcrc32c sd_mod t10_pi ibmvscsi ibmveth scsi_transport_srp
[ 70.856975] CPU: 13 PID: 6480 Comm: insmod Kdump: loaded Tainted: G OE 5.6.0-rc2-ganesh+ #6
[ 70.856985] NIP: c000000000853920 LR: c000000000853a14 CTR: c0000000000376b0
[ 70.856994] REGS: c000000007e4b870 TRAP: 0300 Tainted: G OE (5.6.0-rc2-ganesh+)
[ 70.857003] MSR: 8000000000001003 <SF,ME,RI,LE> CR: 88000422 XER: 00000009
[ 70.857015] CFAR: c000000000853a10 DAR: c0000001ffffa8f8 DSISR: 40000000 IRQMASK: 1
[ 70.857015] GPR00: c000000000853a14 c000000007e4bb00 c000000001372b00 c0000001ffffa8c8
[ 70.857015] GPR04: c000000000cf8728 0000000000000000 0000000000000002 c008000000420810
[ 70.857015] GPR08: 0000000000000000 0000000000000000 0000000000000001 0000000000000001
[ 70.857015] GPR12: 0000000000000000 c000000007f92000 c0000001f8113d70 c00800000059070d
[ 70.857015] GPR16: 00000000000004f8 c008000000421080 000000000000fff1 c008000000421038
[ 70.857015] GPR20: c00000000125eb20 c000000000d1d1c8 c008000000590000 0000000000000000
[ 70.857015] GPR24: 4000000000000510 c008000008000000 c0000000012355d8 c008000000420940
[ 70.857015] GPR28: c008000008000011 0000000000000000 c000000000cf8728 c00000000169a098
[ 70.857097] NIP [c000000000853920] __of_find_property+0x30/0xd0
[ 70.857106] LR [c000000000853a14] of_find_property+0x54/0x90
[ 70.857113] Call Trace:
[ 70.857117] Instruction dump:
[ 70.857124] 3c4c00b2 3842f210 2c230000 418200bc 7c0802a6 fba1ffe8 fbc1fff0 7cbd2b78
[ 70.857136] fbe1fff8 7c9e2378 f8010010 f821ffc1 <ebe30030> 2fbf0000 409e0014 48000064
[ 70.857152] ---[ end trace 13755f7502f3150b ]---
[ 70.864199]
[ 70.864226] Sending IPI to other CPUs
[ 82.011761] ERROR: 15 cpu(s) not responding
</pre>
<blockquote type="cite"
cite="mid:1584157063.g5s75uhbdu.astroid@bobo.none">
<blockquote type="cite">
<pre class="moz-quote-pre" wrap="">This patch fixes this by enabling translation in the exception handler
when all required real mode handling is done. This change only affects
the pSeries platform.
</pre>
</blockquote>
<pre class="moz-quote-pre" wrap="">
Not supposed to do this, because we might not be in a state
where the MMU is ready to be turned on at this point.
I'd like to understand better which accesses are a problem, and whether
we can fix them all to be in the RMO.</pre>
</blockquote>
<pre>I faced three such access problems,
* accessing per-cpu data (like mce_event,mce_event_queue and mce_event_queue),
we can move this inside RMO.
* calling fwnmi_release_errinfo().
* And queuing work to irq_work_queue, not sure how to fix this.
</pre>
<blockquote type="cite"
cite="mid:1584157063.g5s75uhbdu.astroid@bobo.none">
<pre class="moz-quote-pre" wrap="">Thanks,
Nick
</pre>
</blockquote>
<br>
</body>
</html>