Message-ID: <9299ee92-a32b-4b82-aa37-c7087a5c1376@linux.dev>
Date: Thu, 23 May 2024 11:21:52 -0400
From: Sean Anderson <sean.anderson@...ux.dev>
To: Bjorn Helgaas <helgaas@...nel.org>
Cc: Lorenzo Pieralisi <lpieralisi@...nel.org>,
 Krzysztof Wilczyński <kw@...ux.com>,
 Rob Herring <robh@...nel.org>, linux-pci@...r.kernel.org,
 Michal Simek <michal.simek@....com>,
 Thippeswamy Havalige <thippeswamy.havalige@....com>,
 linux-arm-kernel@...ts.infradead.org, Bjorn Helgaas <bhelgaas@...gle.com>,
 linux-kernel@...r.kernel.org, stable@...r.kernel.org,
 Bharat Kumar Gogada <bharatku@...inx.com>
Subject: Re: [PATCH v3 2/7] PCI: xilinx-nwl: Fix off-by-one in IRQ handler

On 5/22/24 18:28, Bjorn Helgaas wrote:
> On Mon, May 20, 2024 at 10:53:57AM -0400, Sean Anderson wrote:
>> MSGF_LEG_MASK is laid out with INTA in bit 0, INTB in bit 1, INTC in bit
>> 2, and INTD in bit 3. Hardware IRQ numbers start at 0, and we register
>> PCI_NUM_INTX irqs. So to enable INTA (aka hwirq 0) we should set bit 0.
>> Remove the subtraction of one. This fixes the following UBSAN error:
> 
> Thanks for these details!
> 
> I guess UBSAN == "undefined behavior sanitizer", right?  That sounds
> like an easy way to find this but not the way users are likely to find
> it.

It's pretty likely they will find it this way, since I found it this way
and no one else had ;)

> I assume users would notice spurious and missing interrupts, e.g.,
> a driver that tried to enable INTB would have actually enabled INTA,
> so we'd see spurious INTA interrupts and the driver would never see
> the INTB it expected.
> 
> And a driver that tried to enable INTA would never see that interrupt,
> and we might not set any bit in MSGF_LEG_MASK?

And yes, this would manifest as INTx interrupts being broken.

> I think the normal way people would trip over this, i.e., spurious and
> missing INTx interrupts, is the important thing to mention here.
> 
>> [    5.037483] ================================================================================
>> [    5.046260] UBSAN: shift-out-of-bounds in ../drivers/pci/controller/pcie-xilinx-nwl.c:389:11
>> [    5.054983] shift exponent 18446744073709551615 is too large for 32-bit type 'int'
>> [    5.062813] CPU: 1 PID: 61 Comm: kworker/u10:1 Not tainted 6.6.20+ #268
>> [    5.070008] Hardware name: xlnx,zynqmp (DT)
>> [    5.074348] Workqueue: events_unbound deferred_probe_work_func
>> [    5.080410] Call trace:
>> [    5.082958] dump_backtrace (arch/arm64/kernel/stacktrace.c:235)
>> [    5.086850] show_stack (arch/arm64/kernel/stacktrace.c:242)
>> [    5.090292] dump_stack_lvl (lib/dump_stack.c:107)
>> [    5.094095] dump_stack (lib/dump_stack.c:114)
>> [    5.097540] __ubsan_handle_shift_out_of_bounds (lib/ubsan.c:218 lib/ubsan.c:387)
>> [    5.103227] nwl_unmask_leg_irq (drivers/pci/controller/pcie-xilinx-nwl.c:389 (discriminator 1))
>> [    5.107386] irq_enable (kernel/irq/internals.h:234 kernel/irq/chip.c:170 kernel/irq/chip.c:439 kernel/irq/chip.c:432 kernel/irq/chip.c:345)
>> [    5.110838] __irq_startup (kernel/irq/internals.h:239 kernel/irq/chip.c:180 kernel/irq/chip.c:250)
>> [    5.114552] irq_startup (kernel/irq/chip.c:270)
>> [    5.118266] __setup_irq (kernel/irq/manage.c:1800)
>> [    5.121982] request_threaded_irq (kernel/irq/manage.c:2206)
>> [    5.126412] pcie_pme_probe (include/linux/interrupt.h:168 drivers/pci/pcie/pme.c:348)
> 
> The rest of the stacktrace below is not relevant and could be omitted.
> The timestamps don't add useful information either.

OK

--Sean

>> [    5.130303] pcie_port_probe_service (drivers/pci/pcie/portdrv.c:528)
>> [    5.134915] really_probe (drivers/base/dd.c:579 drivers/base/dd.c:658)
>> [    5.138720] __driver_probe_device (drivers/base/dd.c:800)
>> [    5.143236] driver_probe_device (drivers/base/dd.c:830)
>> [    5.147571] __device_attach_driver (drivers/base/dd.c:959)
>> [    5.152179] bus_for_each_drv (drivers/base/bus.c:457)
>> [    5.156163] __device_attach (drivers/base/dd.c:1032)
>> [    5.160147] device_initial_probe (drivers/base/dd.c:1080)
>> [    5.164488] bus_probe_device (drivers/base/bus.c:532)
>> [    5.168471] device_add (drivers/base/core.c:3638)
>> [    5.172098] device_register (drivers/base/core.c:3714)
>> [    5.175994] pcie_portdrv_probe (drivers/pci/pcie/portdrv.c:309 drivers/pci/pcie/portdrv.c:363 drivers/pci/pcie/portdrv.c:695)
>> [    5.180338] pci_device_probe (drivers/pci/pci-driver.c:324 drivers/pci/pci-driver.c:392 drivers/pci/pci-driver.c:417 drivers/pci/pci-driver.c:460)
>> [    5.184410] really_probe (drivers/base/dd.c:579 drivers/base/dd.c:658)
>> [    5.188213] __driver_probe_device (drivers/base/dd.c:800)
>> [    5.192729] driver_probe_device (drivers/base/dd.c:830)
>> [    5.197064] __device_attach_driver (drivers/base/dd.c:959)
>> [    5.201672] bus_for_each_drv (drivers/base/bus.c:457)
>> [    5.205657] __device_attach (drivers/base/dd.c:1032)
>> [    5.209641] device_attach (drivers/base/dd.c:1074)
>> [    5.213357] pci_bus_add_device (drivers/pci/bus.c:352)
>> [    5.217518] pci_bus_add_devices (drivers/pci/bus.c:371 (discriminator 2))
>> [    5.221774] pci_host_probe (drivers/pci/probe.c:3099)
>> [    5.225581] nwl_pcie_probe (drivers/pci/controller/pcie-xilinx-nwl.c:938)
>> [    5.229562] platform_probe (drivers/base/platform.c:1404)
>> [    5.233367] really_probe (drivers/base/dd.c:579 drivers/base/dd.c:658)
>> [    5.237169] __driver_probe_device (drivers/base/dd.c:800)
>> [    5.241685] driver_probe_device (drivers/base/dd.c:830)
>> [    5.246020] __device_attach_driver (drivers/base/dd.c:959)
>> [    5.250628] bus_for_each_drv (drivers/base/bus.c:457)
>> [    5.254612] __device_attach (drivers/base/dd.c:1032)
>> [    5.258596] device_initial_probe (drivers/base/dd.c:1080)
>> [    5.262938] bus_probe_device (drivers/base/bus.c:532)
>> [    5.266920] deferred_probe_work_func (drivers/base/dd.c:124)
>> [    5.271619] process_one_work (arch/arm64/include/asm/jump_label.h:21 include/linux/jump_label.h:207 include/trace/events/workqueue.h:108 kernel/workqueue.c:2632)
>> [    5.275788] worker_thread (kernel/workqueue.c:2694 (discriminator 2) kernel/workqueue.c:2781 (discriminator 2))
>> [    5.279686] kthread (kernel/kthread.c:388)
>> [    5.283048] ret_from_fork (arch/arm64/kernel/entry.S:862)
>> [    5.286765] ================================================================================
>> 
>> Fixes: 9a181e1093af ("PCI: xilinx-nwl: Modify IRQ chip for legacy interrupts")
>> Cc: <stable@...r.kernel.org>
>> Signed-off-by: Sean Anderson <sean.anderson@...ux.dev>
>> ---
>> 
>> Changes in v3:
>> - Expand commit message
>> 
>>  drivers/pci/controller/pcie-xilinx-nwl.c | 4 ++--
>>  1 file changed, 2 insertions(+), 2 deletions(-)
>> 
>> diff --git a/drivers/pci/controller/pcie-xilinx-nwl.c b/drivers/pci/controller/pcie-xilinx-nwl.c
>> index 0408f4d612b5..437927e3bcca 100644
>> --- a/drivers/pci/controller/pcie-xilinx-nwl.c
>> +++ b/drivers/pci/controller/pcie-xilinx-nwl.c
>> @@ -371,7 +371,7 @@ static void nwl_mask_intx_irq(struct irq_data *data)
>>  	u32 mask;
>>  	u32 val;
>>  
>> -	mask = 1 << (data->hwirq - 1);
>> +	mask = 1 << data->hwirq;
>>  	raw_spin_lock_irqsave(&pcie->leg_mask_lock, flags);
>>  	val = nwl_bridge_readl(pcie, MSGF_LEG_MASK);
>>  	nwl_bridge_writel(pcie, (val & (~mask)), MSGF_LEG_MASK);
>> @@ -385,7 +385,7 @@ static void nwl_unmask_intx_irq(struct irq_data *data)
>>  	u32 mask;
>>  	u32 val;
>>  
>> -	mask = 1 << (data->hwirq - 1);
>> +	mask = 1 << data->hwirq;
>>  	raw_spin_lock_irqsave(&pcie->leg_mask_lock, flags);
>>  	val = nwl_bridge_readl(pcie, MSGF_LEG_MASK);
>>  	nwl_bridge_writel(pcie, (val | mask), MSGF_LEG_MASK);
>> -- 
>> 2.35.1.1320.gc452695387.dirty
>> 

Powered by blists - more mailing lists