| 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
| ||
|
Date: Thu, 14 Jul 2011 03:42:37 +0530
From: "Koul, Vinod" <vinod.koul@...el.com>
To: Alexander Stein <alexander.stein@...tec-electronic.com>
Cc: Dan Williams <dan.j.williams@...el.com>,
linux-kernel@...r.kernel.org,
Tomoya MORINAGA <tomoya-linux@....okisemi.com>
Subject: Re: [PATCH] pch_dma: Fix channel locking
On Wed, 2011-06-22 at 17:05 +0200, Alexander Stein wrote:
> Fix for the following INFO message
>
> =================================
> [ INFO: inconsistent lock state ]
> 2.6.39+ #89
> ---------------------------------
> inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage.
> rs232/822 [HC1[1]:SC0[0]:HE0:SE1] takes:
> (&(&pd_chan->lock)->rlock){?.....}, at: [<c123b9a1>] pdc_desc_get+0x16/0xab
> {HARDIRQ-ON-W} state was registered at:
> [<c104fe28>] mark_irqflags+0xbd/0x11a
> [<c1050386>] __lock_acquire+0x501/0x6bb
> [<c1050945>] lock_acquire+0x63/0x7b
> [<c131c51d>] _raw_spin_lock_bh+0x43/0x51
> [<c123bee4>] pd_alloc_chan_resources+0x92/0x11e
> [<c123ad62>] dma_chan_get+0x9b/0x107
> [<c123b2d1>] __dma_request_channel+0x61/0xdc
> [<c11ba24b>] pch_request_dma+0x61/0x19e
> [<c11bb3b8>] pch_uart_startup+0x16a/0x1a2
> [<c11b8446>] uart_startup+0x87/0x147
> [<c11b9183>] uart_open+0x117/0x13e
> [<c11a5c7d>] tty_open+0x23c/0x34c
> [<c1097705>] chrdev_open+0x140/0x15f
> [<c10930a6>] __dentry_open.clone.14+0x14a/0x22b
> [<c1093dfb>] nameidata_to_filp+0x36/0x40
> [<c109f28b>] do_last+0x513/0x635
> [<c109f4af>] path_openat+0x9c/0x2aa
> [<c109f6e4>] do_filp_open+0x27/0x69
> [<c1093f02>] do_sys_open+0xfd/0x184
> [<c1093fad>] sys_open+0x24/0x2a
> [<c131d58c>] sysenter_do_call+0x12/0x32
> irq event stamp: 2522
> hardirqs last enabled at (2521): [<c131ca3b>] _raw_spin_unlock_irqrestore+0x36/0x52
> hardirqs last disabled at (2522): [<c131db27>] common_interrupt+0x27/0x34
> softirqs last enabled at (2354): [<c102fa11>] __do_softirq+0x10a/0x11a
> softirqs last disabled at (2299): [<c10041a4>] do_softirq+0x57/0xa4
>
> other info that might help us debug this:
> 2 locks held by rs232/822:
> #0: (&tty->atomic_write_lock){+.+.+.}, at: [<c11a4b7a>] tty_write_lock+0x14/0x3c
> #1: (&port_lock_key){-.....}, at: [<c11bad72>] pch_uart_interrupt+0x17/0x1e9
>
> stack backtrace:
> Pid: 822, comm: rs232 Not tainted 2.6.39+ #89
> Call Trace:
> [<c1319f90>] ? printk+0x19/0x1b
> [<c104f893>] print_usage_bug+0x184/0x18f
> [<c104e5b1>] ? print_irq_inversion_bug+0x10e/0x10e
> [<c104f943>] mark_lock_irq+0xa5/0x1f6
> [<c104fc9c>] mark_lock+0x208/0x2d7
> [<c104fdc0>] mark_irqflags+0x55/0x11a
> [<c1050386>] __lock_acquire+0x501/0x6bb
> [<c10042ee>] ? dump_trace+0x92/0xb6
> [<c1050945>] lock_acquire+0x63/0x7b
> [<c123b9a1>] ? pdc_desc_get+0x16/0xab
> [<c131c2d0>] _raw_spin_lock+0x3e/0x4c
> [<c123b9a1>] ? pdc_desc_get+0x16/0xab
> [<c123b9a1>] pdc_desc_get+0x16/0xab
> [<c10504d8>] ? __lock_acquire+0x653/0x6bb
> [<c123bb2c>] pd_prep_slave_sg+0x7c/0x1cb
> [<c1006c3f>] ? nommu_map_sg+0x6e/0x81
> [<c11bace6>] dma_handle_tx+0x2cf/0x344
> [<c11bad72>] ? pch_uart_interrupt+0x17/0x1e9
> [<c11baebb>] pch_uart_interrupt+0x160/0x1e9
> [<c10642fb>] handle_irq_event_percpu+0x25/0x127
> [<c1064429>] handle_irq_event+0x2c/0x43
> [<c1065e0d>] ? handle_fasteoi_irq+0x84/0x84
> [<c1065eb9>] handle_edge_irq+0xac/0xce
> <IRQ> [<c1003ecb>] ? do_IRQ+0x38/0x9d
> [<c131db2e>] ? common_interrupt+0x2e/0x34
> [<c105007b>] ? __lock_acquire+0x1f6/0x6bb
> [<c131ca3d>] ? _raw_spin_unlock_irqrestore+0x38/0x52
> [<c11b798b>] ? uart_start+0x2d/0x32
> [<c11b7998>] ? uart_flush_chars+0x8/0xa
> [<c11a7962>] ? n_tty_write+0x12c/0x1c6
> [<c1027a73>] ? try_to_wake_up+0x251/0x251
> [<c11a4d0b>] ? tty_write+0x169/0x1dc
> [<c11a7836>] ? n_tty_ioctl+0xb7/0xb7
> [<c1094841>] ? vfs_write+0x91/0x10d
> [<c11a4ba2>] ? tty_write_lock+0x3c/0x3c
> [<c1094a69>] ? sys_write+0x3e/0x63
> [<c131d58c>] ? sysenter_do_call+0x12/0x32
>
> Signed-off-by: Alexander Stein <alexander.stein@...tec-electronic.com>
> ---
> I'm not exactly sure if the dmaengine functions are allowed to be called from
> IRQ context. But I found amba-pl011 which also calls device_prep_slave_sg from
> IRQ context.
>
> drivers/dma/pch_dma.c | 16 ++++++++--------
> 1 files changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/dma/pch_dma.c b/drivers/dma/pch_dma.c
> index ff5b38f..bfb0f19 100644
> --- a/drivers/dma/pch_dma.c
> +++ b/drivers/dma/pch_dma.c
> @@ -495,11 +495,11 @@ static int pd_alloc_chan_resources(struct dma_chan *chan)
> list_add_tail(&desc->desc_node, &tmp_list);
> }
>
> - spin_lock_bh(&pd_chan->lock);
> + spin_lock_irq(&pd_chan->lock);
> list_splice(&tmp_list, &pd_chan->free_list);
> pd_chan->descs_allocated = i;
> pd_chan->completed_cookie = chan->cookie = 1;
> - spin_unlock_bh(&pd_chan->lock);
> + spin_unlock_irq(&pd_chan->lock);
>
> pdc_enable_irq(chan, 1);
>
> @@ -517,10 +517,10 @@ static void pd_free_chan_resources(struct dma_chan *chan)
> BUG_ON(!list_empty(&pd_chan->active_list));
> BUG_ON(!list_empty(&pd_chan->queue));
>
> - spin_lock_bh(&pd_chan->lock);
> + spin_lock_irq(&pd_chan->lock);
> list_splice_init(&pd_chan->free_list, &tmp_list);
> pd_chan->descs_allocated = 0;
> - spin_unlock_bh(&pd_chan->lock);
> + spin_unlock_irq(&pd_chan->lock);
>
> list_for_each_entry_safe(desc, _d, &tmp_list, desc_node)
> pci_pool_free(pd->pool, desc, desc->txd.phys);
> @@ -536,10 +536,10 @@ static enum dma_status pd_tx_status(struct dma_chan *chan, dma_cookie_t cookie,
> dma_cookie_t last_completed;
> int ret;
>
> - spin_lock_bh(&pd_chan->lock);
> + spin_lock_irq(&pd_chan->lock);
> last_completed = pd_chan->completed_cookie;
> last_used = chan->cookie;
> - spin_unlock_bh(&pd_chan->lock);
> + spin_unlock_irq(&pd_chan->lock);
>
> ret = dma_async_is_complete(cookie, last_completed, last_used);
>
> @@ -654,7 +654,7 @@ static int pd_device_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd,
> if (cmd != DMA_TERMINATE_ALL)
> return -ENXIO;
>
> - spin_lock_bh(&pd_chan->lock);
> + spin_lock_irq(&pd_chan->lock);
>
> pdc_set_mode(&pd_chan->chan, DMA_CTL0_DISABLE);
>
> @@ -664,7 +664,7 @@ static int pd_device_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd,
> list_for_each_entry_safe(desc, _d, &list, desc_node)
> pdc_chain_complete(pd_chan, desc);
>
> - spin_unlock_bh(&pd_chan->lock);
> + spin_unlock_irq(&pd_chan->lock);
>
> return 0;
> }
Thanks applied with tested-by from Tomoya
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists