[PATCH 3/4 v1] mtd: spi-nor: fsl-quadspi: fix qspi irq handler complete exception
Han Xu
xhnjupt at gmail.com
Tue Jun 2 21:42:15 PDT 2015
On Mon, May 25, 2015 at 6:50 AM, Haikun Wang <haikun.wang at freescale.com> wrote:
> In case of interrupt arrive immediately after requesting irq,
> kernel will panic due to uninitialized variable.
>
> Fix below exception on LS1021AQDS:
> Unable to handle kernel NULL pointer dereference at virtual address
> 00000000
> pgd = 80003000
> [00000000] *pgd=80000080004003, *pmd=00000000
> Internal error: Oops: 206 [#1] SMP THUMB2
> Modules linked in:
> CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.1.0-rc5+ #2
> Hardware name: Freescale LS1021A
> task: 80526f30 ti: 80522000 task.ti: 80522000
> PC is at __wake_up_common+0x14/0x46
> LR is at __wake_up_locked+0xb/0x10
> pc : [<800388ae>] lr : [<800388eb>] psr: 400001b3
> sp : 80523e78 ip : 00000008 fp : 00000001
> r10: 00000000 r9 : 00000003 r8 : ee008000
> r7 : 00000001 r6 : 80000193 r5 : 00000000 r4 : ee0c0bd0
> r3 : 00000000 r2 : 00000001 r1 : 00000003 r0 : fffffff4
> Flags: nZcv IRQs off FIQs on Mode SVC_32 ISA Thumb Segment kernel
> Control: 70c5387d Table: 80003000 DAC: fddff9d7
> Process swapper/0 (pid: 0, stack limit = 0x80522210)
> Stack: (0x80523e78 to 0x80524000)
> 3e60: 00000000
> ee0c0bcc
> 3e80: ee0c0bc8 80000193 00000000 ee0cb480 805532ff 800388eb 00000000
> 00000003
> 3ea0: 00000001 80038d7f ee0c0010 08010001 00000000 80211587 80211553
> cf81af00
> 3ec0: 00000017 8003e199 ee0cb480 cf81af00 ee0cb480 80532ec0 6e2b4000
> 00000000
> 3ee0: ee008000 00000001 803326d0 8003e265 00000000 ee0cb480 80532ec0
> 8003ff6b
> 3f00: 8003ff09 00000017 8051ebc4 8003dc77 00000000 8003de47 80523f48
> f0002000
> 3f20: 80523f48 805247d4 80523f7c 8051d384 8051fca0 8000924b 8000db28
> 40000133
> 3f40: ffffffff 8032df1b 00000001 00000000 00000000 80017be1 80522000
> 00000000
> 3f60: 00000000 80523f98 8051d384 8051fca0 803326d0 00000001 00000008
> 80523f90
> 3f80: 8000db27 8000db28 40000133 ffffffff 00000000 80038fd7 00000001
> 00000000
> 3fa0: 00000002 80556000 00000000 804eb933 ffffffff ffffffff 804eb545
> 00000000
> 3fc0: ffffffff 00000000 00000000 80512590 00000000 80556294 80524460
> 8051258c
> 3fe0: 80528050 80003010 410fc075 00000000 00000000 8000808f 00000000
> 00000000
> [<800388ae>] (__wake_up_common) from [<800388eb>]
> (__wake_up_locked+0xb/0x10)
> [<800388eb>] (__wake_up_locked) from [<80038d7f>] (complete+0x1f/0x2a)
> [<80038d7f>] (complete) from [<80211587>] (fsl_qspi_irq_handler+0x35/0x38)
> [<80211587>] (fsl_qspi_irq_handler) from [<8003e199>]
> (handle_irq_event_percpu+0x1b/0xb6)
> [<8003e199>] (handle_irq_event_percpu) from [<8003e265>]
> (handle_irq_event+0x31/0x48)
> [<8003e265>] (handle_irq_event) from [<8003ff6b>]
> (handle_fasteoi_irq+0x63/0xb8)
> [<8003ff6b>] (handle_fasteoi_irq) from [<8003dc77>]
> (generic_handle_irq+0x13/0x1c)
> [<8003dc77>] (generic_handle_irq) from [<8003de47>]
> (__handle_domain_irq+0x53/0x74)
> [<8003de47>] (__handle_domain_irq) from [<8000924b>]
> (gic_handle_irq+0x27/0x40)
> [<8000924b>] (gic_handle_irq) from [<8032df1b>] (__irq_svc+0x3b/0x5c)
> Exception stack(0x80523f48 to 0x80523f90)
> 3f40: 00000001 00000000 00000000 80017be1 80522000
> 00000000
> 3f60: 00000000 80523f98 8051d384 8051fca0 803326d0 00000001 00000008
> 80523f90
> 3f80: 8000db27 8000db28 40000133 ffffffff
> [<8032df1b>] (__irq_svc) from [<8000db28>] (arch_cpu_idle+0x14/0x20)
> [<8000db28>] (arch_cpu_idle) from [<80038fd7>]
> (cpu_startup_entry+0x187/0x1c4)
> [<80038fd7>] (cpu_startup_entry) from [<804eb933>]
> (start_kernel+0x27b/0x2e4)
> [<804eb933>] (start_kernel) from [<8000808f>] (0x8000808f)
> Code: 4617 469a f1a5 000c (682d) 3d0c
> ---[ end trace 420f2b58e4270f57 ]---
> Kernel panic - not syncing: Fatal exception in interrupt
> CPU1: stopping
> CPU: 1 PID: 0 Comm: swapper/1 Tainted: G D 4.1.0-rc5+ #2
> Hardware name: Freescale LS1021A
> [<80011937>] (unwind_backtrace) from [<8000f4e7>] (show_stack+0xb/0xc)
> [<8000f4e7>] (show_stack) from [<8032ad55>] (dump_stack+0x51/0x64)
> [<8032ad55>] (dump_stack) from [<80010ff9>] (handle_IPI+0x7d/0xf8)
> [<80010ff9>] (handle_IPI) from [<8000925b>] (gic_handle_irq+0x37/0x40)
> [<8000925b>] (gic_handle_irq) from [<8032df1b>] (__irq_svc+0x3b/0x5c)
> Exception stack(0xee079f90 to 0xee079fd8)
> 9f80: 00000001 00000000 00000000
> 80017be1
> 9fa0: ee078000 00000000 00000000 ee079fe0 8051d384 8051fca0 803326d0
> 00000001
> 9fc0: 00000008 ee079fd8 8000db27 8000db28 40000133 ffffffff
> [<8032df1b>] (__irq_svc) from [<8000db28>] (arch_cpu_idle+0x14/0x20)
> [<8000db28>] (arch_cpu_idle) from [<80038fd7>]
> (cpu_startup_entry+0x187/0x1c4)
> [<80038fd7>] (cpu_startup_entry) from [<80009311>] (__enable_mmu+0x1/0x10)
> ---[ end Kernel panic - not syncing: Fatal exception in interrupt
>
> Signed-off-by: Haikun Wang <haikun.wang at freescale.com>
> ---
> drivers/mtd/spi-nor/fsl-quadspi.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/mtd/spi-nor/fsl-quadspi.c b/drivers/mtd/spi-nor/fsl-quadspi.c
> index 997eb20..fa615a0 100644
> --- a/drivers/mtd/spi-nor/fsl-quadspi.c
> +++ b/drivers/mtd/spi-nor/fsl-quadspi.c
> @@ -666,6 +666,10 @@ static int fsl_qspi_nor_setup(struct fsl_qspi *q)
> qspi_writel(q, QUADSPI_MCR_RESERVED_MASK | QUADSPI_MCR_END_CFG_MASK,
> base + QUADSPI_MCR);
>
> + /* clear flag register before enable the interrupt */
> + reg = qspi_readl(q, q->iobase + QUADSPI_FR);
> + qspi_writel(q, reg, q->iobase + QUADSPI_FR);
> +
I think clear the interrupt before enable interrupt can fix this
issue, init_completion in early stage is not necessary.
Han Xu
> /* enable the interrupt */
> qspi_writel(q, QUADSPI_RSER_TFIE, q->iobase + QUADSPI_RSER);
>
> @@ -878,6 +882,10 @@ static int fsl_qspi_probe(struct platform_device *pdev)
> goto irq_failed;
> }
>
> + /* In case of interrupt arrive immediately after requesting irq,
> + * kernel will panic without init_completion.
> + */
> + init_completion(&q->c);
> ret = devm_request_irq(dev, ret,
> fsl_qspi_irq_handler, 0, pdev->name, q);
> if (ret) {
> --
> 2.1.0.27.g96db324
>
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
More information about the linux-mtd
mailing list