linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Sven Schnelle <svens@linux.ibm.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: linux-kernel@vger.kernel.org, linux-s390@vger.kernel.org,
	hca@linux.ibm.com
Subject: crash on s390 with SCHED_CORE=y
Date: Thu, 26 Aug 2021 11:06:31 +0200	[thread overview]
Message-ID: <yt9d35qwa82w.fsf@linux.ibm.com> (raw)

Hi Peter,

i'm seeing the following crash on s390 with the strace testsuite (namely
the prctl-sched-core--pidns-translation testcase).

[   25.044039] Unable to handle kernel pointer dereference in virtual kernel address space
[   25.044051] Failing address: 0000000000000000 TEID: 0000000000000483
[   25.044053] Fault in home space mode while using kernel ASCE.
[   25.044056] AS:0000000166bd4007 R3:00000003fffe8007 S:00000003fffed800 P:000000000000003d
[   25.044080] Oops: 0004 ilc:2 [#1] SMP
[   25.044085] Modules linked in: nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ip_set nf_tables nfnetlink sunrpc dm_multipath scsi_dh_rdac scsi_dh_emc scsi_dh_alua s390_trng ghash_s390 ism p
rng smc aes_s390 des_s390 libdes ib_core sha3_512_s390 sha3_256_s390 sha512_s390 sha256_s390 sha1_s390 sha_common eadm_sch vfio_ccw mdev vfio_iommu_type1 vfio sch_fq_codel pkey zcrypt rng_core configfs ip_tables x_tables autofs4
[   25.044125] CPU: 122 PID: 2045 Comm: prctl-sched-cor Not tainted 5.14.0-rc7-08025-gf6d7568b37df-dirty #532
[   25.044129] Hardware name: IBM 8561 T01 701 (LPAR)
[   25.044130] Krnl PSW : 0404e00180000000 00000001655b7534 (raw_spin_rq_lock_nested+0x5c/0xb8)
[   25.044142]            R:0 T:1 IO:0 EX:0 Key:0 M:1 W:0 P:0 AS:3 CC:2 PM:0 RI:0 EA:3
[   25.044145] Krnl GPRS: ffffffffffffffff 0000000000000000 000000030000007b 0000000000000000
[   25.044147]            0000000000000000 0000000000000000 0000000100000080 0000000166ad0900
[   25.044149]            0000000000000400 0000000100000000 0000000000000000 00000003fd9a3600
[   25.044150]            000000009fb50000 0000000166172538 00000380077cbd00 00000380077cbcc0
[   25.044158] Krnl Code: 00000001655b752a: b90400ab            lgr     %r10,%r11
[   25.044158]            00000001655b752e: 582003ac            l       %r2,940
[   25.044158]           #00000001655b7532: 1819                lr      %r1,%r9
[   25.044158]           >00000001655b7534: ba12a000            cs      %r1,%r2,0(%r10)
[   25.044158]            00000001655b7538: ec16001d007e        cij     %r1,0,6,00000001655b7572
[   25.044158]            00000001655b753e: e310b1300112        lt      %r1,4400(%r11)
[   25.044158]            00000001655b7544: a7740012            brc     7,00000001655b7568
[   25.044158]            00000001655b7548: b904001b            lgr     %r1,%r11
[   25.044230] Call Trace:
[   25.044234]  [<00000001655b7534>] raw_spin_rq_lock_nested+0x5c/0xb8
[   25.044241]  [<00000001655d2cfc>] online_fair_sched_group+0x9c/0x1c0
[   25.044248]  [<00000001655e481c>] sched_autogroup_create_attach+0xdc/0x210
[   25.044253]  [<000000016559a45e>] ksys_setsid+0x10e/0x148
[   25.044258]  [<000000016559a4ba>] __s390_sys_setsid+0x22/0x30
[   25.044261]  [<000000016608402c>] __do_syscall+0x1bc/0x1e8
[   25.044266]  [<0000000166091358>] system_call+0x78/0xa0
[   25.044269] Last Breaking-Event-Address:
[   25.044270]  [<0000000000000000>] 0x0
[   25.044274] Kernel panic - not syncing: Fatal exception: panic_on_oops

The crash is caused by using a NULL rq->core value in
__rq_lockp, which is then passed on to raw_spin_rq_lock_nested(). This
is because it is implemented as follows:

static inline raw_spinlock_t *__rq_lockp(struct rq *rq)
{
        if (rq->core_enabled)
		return &rq->core->__lock;
	return &rq->__lock;
}

The problem is, that __sched_core_flip() sets rq->core_enabled to 1 even
for offline CPUS:

	/*
	 * Toggle the offline CPUs.
	 */
	cpumask_copy(&sched_core_mask, cpu_possible_mask);
	cpumask_andnot(&sched_core_mask, &sched_core_mask, cpu_online_mask);

	for_each_cpu(cpu, &sched_core_mask)
		cpu_rq(cpu)->core_enabled = enabled;

... But rq->core seems to get only assigned in the cpu hotplug notifer
sched_core_cpu_starting(), which is never called for offline CPUs. At
least i haven't found another place which assigns rq->core.

Any thoughts?

Thanks
Svem

             reply	other threads:[~2021-08-26  9:06 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-08-26  9:06 Sven Schnelle [this message]
2021-08-26 10:08 ` crash on s390 with SCHED_CORE=y Peter Zijlstra
2021-08-26 10:21   ` Sven Schnelle

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=yt9d35qwa82w.fsf@linux.ibm.com \
    --to=svens@linux.ibm.com \
    --cc=hca@linux.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-s390@vger.kernel.org \
    --cc=peterz@infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).