Skip to content

R
raspberrypi-kernel

openeuler / raspberrypi-kernel

通知 13
Star 1
Fork 0
R
raspberrypi-kernel
切换分支/标签
  1. 16 2月, 2015 1 次提交
  3. 12 2月, 2015 3 次提交
  5. 11 2月, 2015 1 次提交
    • A
      dmaengine: pl330: fix bug that cause start the same descs in cyclic · 0091b9d6
      由 Addy Ke 提交于 
      This bug will cause NULL pointer after commit dfac17, and cause
wrong package in I2S DMA transfer before commit dfac17.

Tested on RK3288-pinky2 board.

Detail:
I2S DMA transfer(sound/core/pcm_dmaengine.c):
dmaengine_pcm_prepare_and_submit -->
dmaengine_prep_dma_cyclic -->
pl330_prep_dma_cyclic -->
the case:
1. pl330_submit_req(desc0): thrd->req[0].desc = desc0, thrd->lstenq = 0
2. pl330_submit_req(desc1): thrd->req[1].desc = desc1, thrd->lstenq = 1
3. _start(desc0) by submit_req: thrd->req_running = 0
   because: idx = 1 - thrd->lstenq = 0
4. pl330_update(desc0 OK): thrd->req[0].desc = NULL, desc0 to req_done list
   because: idx = active = thrd->req_running = 0
5. _start(desc1) by pl330_update: thrd->req_running = 1
   because:
   idx = 1 - thrd->lstenq = 0, but thrd->req[0].desc == NULL,
   so:
   idx = thrd->lstenq = 1
6. pl330_submit_req(desc2): thrd->req[0].desc = desc2, thrd->lstenq = 0
7. _start(desc1) by submit_req: thrd->req_running = 1
   because: idx = 1 - thrd->lstenq = 1
   Note: _start started the same descs
         _start should start desc2 here, NOT desc1

8. pl330_update(desc1 OK): thrd->req[1].desc = NULL, desc1 to req_done list
   because: idx = active = thrd->req_running = 1
9. _start(desc2) by pl330_update : thrd->req_running = 0
   because: idx = 1 - thrd->lstenq = 0
10.pl330_update(desc1 OK, NOT desc2): thrd->req[0].desc = NULL,
   desc2 to req_done list
   because: idx = active = thrd->req_running = 0

11.pl330_submit_req(desc3): thrd->req[0].desc = desc3, thrd->lstenq = 0
12.pl330_submit_req(desc4): thrd->req[1].desc = desc4, thrd->lstenq = 1
13._start(desc3) by submit_req: thrd->req_running = 0
   because: idx = 1 - thrd->lstenq = 0
14.pl330_update(desc2 OK NOT desc3): thrd->req[0].desc = NULL
   desc3 to req_done list
   because: idx = active = thrd->req_running = 0
15._start(desc4) by pl330_update: thrd->req_running = 1
   because:
   idx = 1 - thrd->lstenq = 0, but thrd->req[0].desc == NULL,
   so:
   idx = thrd->lstenq = 1
16.pl330_submit_req(desc5): thrd->req[0].desc = desc5, thrd->lstenq = 0
17._start(desc4) by submit_req: thrd->req_running = 1
   because: idx = 1 - thrd->lstenq = 1
18.pl330_update(desc3 OK NOT desc4): thrd->req[1].desc = NULL
   desc4 to req_done list
   because: idx = active = thrd->req_running = 1
19._start(desc4) by pl330_update: thrd->req_running = 0
   because:
   idx = 1 - thrd->lstenq = 1, but thrd->req[1].desc == NULL,
   so:
   idx = thrd->lstenq = 0
20.pl330_update(desc4 OK): thrd->req[0].desc = NULL, desc5 to req_done list
   because: idx = active = thrd->req_running = 0
21.pl330_update(desc4 OK):
   1) before commit dfac17(set req_running -1 in pl330_update/mark_free()):
      because: active = -1, abort
      result: desc0-desc5's callback are all called,
	      but step 10 and step 18 go wrong.
   2) before commit dfac17:
      idx = active = thrd->req_runnig = 0 -->
      descdone = thrd->req[0] = NULL -->
      list_add_tail(&descdone->rqd, &pl330->req_done); -->
      got NULL pointer!!!
Signed-off-by: NAddy Ke <addy.ke@rock-chips.com>
Signed-off-by: NVinod Koul <vinod.koul@intel.com>
      0091b9d6
  7. 05 2月, 2015 10 次提交
  9. 04 2月, 2015 1 次提交
  11. 26 1月, 2015 3 次提交
  13. 18 1月, 2015 4 次提交
  15. 14 1月, 2015 11 次提交
  17. 13 1月, 2015 4 次提交
  19. 23 12月, 2014 2 次提交
    • J
      dmaengine: rcar-dmac: Handle hardware descriptor allocation failure · ee4b876b
      由 Jürg Billeter 提交于 
      If the atomic DMA coherent pool is too small, disable use of hardware
descriptor lists instead of crashing the system:

ERROR: 256 KiB atomic DMA coherent pool is too small!
Please increase it with coherent_pool= kernel parameter!

Unable to handle kernel NULL pointer dereference at virtual address 00000004
Internal error: Oops: a07 [#1] PREEMPT SMP ARM

PC is at rcar_dmac_chan_reinit+0x3c/0x160
LR is at _raw_spin_lock_irqsave+0x18/0x5c

[<802132c0>] (rcar_dmac_chan_reinit) from [<80214818>] (rcar_dmac_isr_error+0x84/0xa0)
[<80214818>] (rcar_dmac_isr_error) from [<80060484>] (handle_irq_event_percpu+0x50/0x150)
[<80060484>] (handle_irq_event_percpu) from [<800605c0>] (handle_irq_event+0x3c/0x5c)
[<800605c0>] (handle_irq_event) from [<8006350c>] (handle_fasteoi_irq+0xb8/0x198)
[<8006350c>] (handle_fasteoi_irq) from [<8005fdb0>] (generic_handle_irq+0x20/0x30)
[<8005fdb0>] (generic_handle_irq) from [<8000fcd0>] (handle_IRQ+0x50/0xc4)
[<8000fcd0>] (handle_IRQ) from [<800092cc>] (gic_handle_irq+0x28/0x5c)
[<800092cc>] (gic_handle_irq) from [<80012700>] (__irq_svc+0x40/0x70)

Kernel panic - not syncing: Fatal exception in interrupt
Signed-off-by: NJürg Billeter <j@bitron.ch>
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
      ee4b876b
    • L
      dmaengine: rcar-dmac: Cache hardware descriptors memory · 1ed1315f
      由 Laurent Pinchart 提交于 
      Unlike DMA transfers descriptors that are preallocated and cached,
memory used to store hardware descriptors is allocated and freed with
the DMA coherent allocation API for every transfer. Besides degrading
performances, this creates a CMA stress test that seems to cause issues.
Running dmatest with the noverify option produces

[   50.066539] alloc_contig_range test_pages_isolated(6b845, 6b846) failed
[   50.235180] alloc_contig_range test_pages_isolated(6b848, 6b84e) failed
[   52.964584] alloc_contig_range test_pages_isolated(6b847, 6b848) failed
[   54.127113] alloc_contig_range test_pages_isolated(6b843, 6b844) failed
[   56.270253] alloc_contig_range test_pages_isolated(6b84c, 6b850) failed

The root cause needs to be fixed, but in the meantime, as a workaround
and a performance improvement, cache hardware descriptors.
Signed-off-by: NLaurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Tested-by: NWolfram Sang <wsa+renesas@sang-engineering.com>
      1ed1315f