linux-kernel - Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible

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>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <2a0657bf-2585-4b98-bbdb-dd38502c8555@csgroup.eu>
Date:   Tue, 12 Sep 2023 15:06:27 +0000
From:   Christophe Leroy <christophe.leroy@...roup.eu>
To:     "paulmck@...nel.org" <paulmck@...nel.org>,
        "Liam R. Howlett" <Liam.Howlett@...cle.com>,
        Geert Uytterhoeven <geert@...ux-m68k.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        "maple-tree@...ts.infradead.org" <maple-tree@...ts.infradead.org>,
        "linux-mm@...ck.org" <linux-mm@...ck.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "stable@...r.kernel.org" <stable@...r.kernel.org>,
        "linux-renesas-soc@...r.kernel.org" 
        <linux-renesas-soc@...r.kernel.org>,
        Shanker Donthineni <sdonthineni@...dia.com>
Subject: Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other
 readers are possible



Le 12/09/2023 à 16:37, Christophe Leroy a écrit :
> 
> 
>>
>> Maybe something like this in __setup_irq(), right before the
>> mutex_lock()?
>>
>> 	WARN_ON_ONCE(irqs_disabled());
>>
>> This will dump the stack trace showing how __setup_irq() is being invoked
>> in early boot on ppc32.
>>
>> Again, given that __setup_irq() invokes mutex_lock(), invoking this
>> function in its current form before interrupts have been enabled is a
>> bad idea.
>>
> 
> No trigger of that WARN_ON() I added in __setup_irq() as instructed
> above, still getting (pmac32_defconfig on MAC99 QEMU)
> 
> ------------[ cut here ]------------
> Interrupts were enabled early
> WARNING: CPU: 0 PID: 0 at init/main.c:992 start_kernel+0x4d8/0x5c0
> Modules linked in:
> CPU: 0 PID: 0 Comm: swapper Not tainted 6.6.0-rc1-dirty #481
> Hardware name: PowerMac3,1 7400 0xc0209 PowerMac
> NIP:  c0a6052c LR: c0a6052c CTR: 00000000
> REGS: c0c4dee0 TRAP: 0700   Not tainted  (6.6.0-rc1-dirty)
> MSR:  00029032 <EE,ME,IR,DR,RI>  CR: 24000282  XER: 20000000
> 
> GPR00: c0a6052c c0c4dfa0 c0b92580 0000001d c0b9d128 00000001 c0b9d148
> 3ffffdff
> GPR08: c0ba80f0 00000000 00000000 3ffffe00 44000282 00000000 00000000
> 0199abfc
> GPR16: 0199b0a4 7fde7fa4 7fc5ac0c 000000bb 41000000 01c690c8 c0b92014
> c09b4bdc
> GPR24: c0c55220 c0ac0000 00000000 efff9109 efff9100 0000000a c0c6d000
> c0b920a0
> NIP [c0a6052c] start_kernel+0x4d8/0x5c0
> LR [c0a6052c] start_kernel+0x4d8/0x5c0
> Call Trace:
> [c0c4dfa0] [c0a6052c] start_kernel+0x4d8/0x5c0 (unreliable)
> [c0c4dff0] [00003540] 0x3540
> Code: 480037b1 48023c05 4bab88ed 90620260 480139e9 4b657ced 7d2000a6
> 71298000 41a20014 3c60c09a 3863b78c 4b5e9ccd <0fe00000> 39200000
> 99380008 7d2000a6
> ---[ end trace 0000000000000000 ]---
> 

For what it's worth, the interrupts seems to be enabled by the call to 
init_IRQ().

Diging into it that's enabled by the call to __vmalloc_node() in 
alloc_vm_stack()

static void *__init alloc_vm_stack(void)
{
	return __vmalloc_node(THREAD_SIZE, THREAD_ALIGN, THREADINFO_GFP,
			      NUMA_NO_NODE, (void *)_RET_IP_);
}

static void __init vmap_irqstack_init(void)
{
	int i;

	for_each_possible_cpu(i) {
		softirq_ctx[i] = alloc_vm_stack();
		hardirq_ctx[i] = alloc_vm_stack();
	}
}


void __init init_IRQ(void)
{
	if (IS_ENABLED(CONFIG_VMAP_STACK))
		vmap_irqstack_init();

	if (ppc_md.init_IRQ)
		ppc_md.init_IRQ();

	if (!WARN_ON(!ppc_md.get_irq))
		static_call_update(ppc_get_irq, ppc_md.get_irq);
}

Christophe