lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Date:   Wed, 1 Nov 2017 16:45:56 +0000
From:   Matt Redfearn <matt.redfearn@...s.com>
To:     Ralf Baechle <ralf@...ux-mips.org>, James Hogan <jhogan@...nel.org>
CC:     Matt Redfearn <matt.redfearn@...s.com>,
        <linux-mips@...ux-mips.org>, "Paul Burton" <paul.burton@...s.com>,
        <linux-kernel@...r.kernel.org>, Ingo Molnar <mingo@...nel.org>,
        Paul Burton <paul.burton@...tec.com>
Subject: [PATCH] MIPS: CPS: Fix use of current_cpu_data in preemptible code

Commit 1ec9dd80bedc2 ("MIPS: CPS: Detect CPUs in secondary clusters")
added a check in cps_boot_secondary() that the secondary being booted is
in the same cluster as the CPU running this code. This check is
performed using current_cpu_data without disabling preemption. As such
when CONFIG_PREEMPT=y, a BUG is triggered:

[   57.991693] BUG: using smp_processor_id() in preemptible [00000000] code: hotplug/1749
<snip>
[   58.063077] Call Trace:
[   58.065842] [<8040cdb4>] show_stack+0x84/0x114
[   58.070830] [<80b11b38>] dump_stack+0xf8/0x140
[   58.075796] [<8079b12c>] check_preemption_disabled+0xec/0x118
[   58.082204] [<80415110>] cps_boot_secondary+0x84/0x44c
[   58.087935] [<80413a14>] __cpu_up+0x34/0x98
[   58.092624] [<80434240>] bringup_cpu+0x38/0x114
[   58.097680] [<80434af0>] cpuhp_invoke_callback+0x168/0x8f0
[   58.103801] [<804362d0>] _cpu_up+0x154/0x1c8
[   58.108565] [<804363dc>] do_cpu_up+0x98/0xa8
[   58.113333] [<808261f8>] device_online+0x84/0xc0
[   58.118481] [<80826294>] online_store+0x60/0x98
[   58.123562] [<8062261c>] kernfs_fop_write+0x158/0x1d4
[   58.129196] [<805a2ae4>] __vfs_write+0x4c/0x168
[   58.134247] [<805a2dc8>] vfs_write+0xe0/0x190
[   58.139095] [<805a2fe0>] SyS_write+0x68/0xc4
[   58.143854] [<80415d58>] syscall_common+0x34/0x58

In reality we don't currently support running the kernel on CPUs not in
cluster 0, so the answer to cpu_cluster(&current_cpu_data) will always
be 0, even if this task being preempted and continues running on a
different CPU. Regardless, the BUG should not be triggered, so fix this
by switching to raw_current_cpu_data. When multicluster support lands
upstream this check will need removing or changing anyway.

Signed-off-by: Matt Redfearn <matt.redfearn@...s.com>
Reviewed-by: Paul Burton <paul.burton@...s.com>
CC: linux-mips@...ux-mips.org
CC: Paul Burton <paul.burton@...s.com>
---

 arch/mips/kernel/smp-cps.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/mips/kernel/smp-cps.c b/arch/mips/kernel/smp-cps.c
index 0063122c85da..8e7a40bcd9d3 100644
--- a/arch/mips/kernel/smp-cps.c
+++ b/arch/mips/kernel/smp-cps.c
@@ -306,7 +306,7 @@ static int cps_boot_secondary(int cpu, struct task_struct *idle)
 	int err;
 
 	/* We don't yet support booting CPUs in other clusters */
-	if (cpu_cluster(&cpu_data[cpu]) != cpu_cluster(&current_cpu_data))
+	if (cpu_cluster(&cpu_data[cpu]) != cpu_cluster(&raw_current_cpu_data))
 		return -ENOSYS;
 
 	vpe_cfg->pc = (unsigned long)&smp_bootstrap;
-- 
2.7.4

Powered by blists - more mailing lists