On Wed, Jun 03, 2015 at 04:52:26PM +0200, Ludovic Desroches wrote:
> Using _bh variant for spin locks causes this kind of warning:
> Starting logging: ------------[ cut here ]------------
> WARNING: CPU: 0 PID: 3 at /ssd_drive/linux/kernel/softirq.c:151
> __local_bh_enable_ip+0xe8/0xf4()
> Modules linked in:
> CPU: 0 PID: 3 Comm: ksoftirqd/0 Not tainted 4.1.0-rc2+ #94
> Hardware name: Atmel SAMA5
> [<c0013c04>] (unwind_backtrace) from [<c00118a4>] (show_stack+0x10/0x14)
> [<c00118a4>] (show_stack) from [<c001bbcc>]
> (warn_slowpath_common+0x80/0xac)
> [<c001bbcc>] (warn_slowpath_common) from [<c001bc14>]
> (warn_slowpath_null+0x1c/0x24)
> [<c001bc14>] (warn_slowpath_null) from [<c001e28c>]
> (__local_bh_enable_ip+0xe8/0xf4)
> [<c001e28c>] (__local_bh_enable_ip) from [<c01fdbd0>]
> (at_xdmac_device_terminate_all+0xf4/0x100)
> [<c01fdbd0>] (at_xdmac_device_terminate_all) from [<c02221a4>]
> (atmel_complete_tx_dma+0x34/0xf4)
> [<c02221a4>] (atmel_complete_tx_dma) from [<c01fe4ac>]
> (at_xdmac_tasklet+0x14c/0x1ac)
> [<c01fe4ac>] (at_xdmac_tasklet) from [<c001de58>]
> (tasklet_action+0x68/0xb4)
> [<c001de58>] (tasklet_action) from [<c001dfdc>]
> (__do_softirq+0xfc/0x238)
> [<c001dfdc>] (__do_softirq) from [<c001e140>] (run_ksoftirqd+0x28/0x34)
> [<c001e140>] (run_ksoftirqd) from [<c0033a3c>]
> (smpboot_thread_fn+0x138/0x18c)
> [<c0033a3c>] (smpboot_thread_fn) from [<c0030e7c>] (kthread+0xdc/0xf0)
> [<c0030e7c>] (kthread) from [<c000f480>] (ret_from_fork+0x14/0x34)
> ---[ end trace b57b14a99c1d8812 ]---
>
> It comes from the fact that devices can called some code from the DMA
> controller with irq disabled. _bh variant is not intended to be used in
> this case since it can enable irqs. Switch to irqsave/irqrestore variant to
> avoid this situation.
>
> Signed-off-by: Ludovic Desroches <[email protected]>
> Cc: [email protected] # 4.0 and later
> ---
> drivers/dma/at_xdmac.c | 85
> ++++++++++++++++++++++++++++----------------------
> 1 file changed, 48 insertions(+), 37 deletions(-)
>
> diff --git a/drivers/dma/at_xdmac.c b/drivers/dma/at_xdmac.c
> index 9b602a6..4a7e9c6 100644
> --- a/drivers/dma/at_xdmac.c
> +++ b/drivers/dma/at_xdmac.c
> @@ -421,8 +421,9 @@ static dma_cookie_t at_xdmac_tx_submit(struct
> dma_async_tx_descriptor *tx)
> struct at_xdmac_desc *desc = txd_to_at_desc(tx);
> struct at_xdmac_chan *atchan = to_at_xdmac_chan(tx->chan);
> dma_cookie_t cookie;
> + unsigned long irqflags;
>
> - spin_lock_bh(&atchan->lock);
> + spin_lock_irqsave(&atchan->lock, irqflags);
> cookie = dma_cookie_assign(tx);
>
> dev_vdbg(chan2dev(tx->chan), "%s: atchan 0x%p, add desc 0x%p to
> xfers_list\n",
> @@ -431,7 +432,7 @@ static dma_cookie_t at_xdmac_tx_submit(struct
> dma_async_tx_descriptor *tx)
> if (list_is_singular(&atchan->xfers_list))
> at_xdmac_start_xfer(atchan, desc);
>
> - spin_unlock_bh(&atchan->lock);
> + spin_unlock_irqrestore(&atchan->lock, irqflags);
> return cookie;
> }
>
> @@ -591,11 +592,13 @@ at_xdmac_prep_slave_sg(struct dma_chan *chan, struct
> scatterlist *sgl,
> unsigned int sg_len, enum dma_transfer_direction
> direction,
> unsigned long flags, void *context)
> {
> - struct at_xdmac_chan *atchan = to_at_xdmac_chan(chan);
> - struct at_xdmac_desc *first = NULL, *prev = NULL;
> - struct scatterlist *sg;
> - int i;
> - unsigned int xfer_size = 0;
> + struct at_xdmac_chan *atchan = to_at_xdmac_chan(chan);
> + struct at_xdmac_desc *first = NULL, *prev = NULL;
> + struct scatterlist *sg;
> + int i;
> + unsigned int xfer_size = 0;
> + unsigned long irqflags;
> + struct dma_async_tx_descriptor *ret = NULL;
This doesnt comply with stable rules, and you should not keep this in fixes
either :(
- It cannot contain any "trivial" fixes in it (spelling changes,
whitespace cleanups, etc).
--
~Vinod
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
