This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Linux/DRA77P: DRA7XXP bb2d support

Part Number: DRA77P

Tool/software: Linux

Hello all

I try to get GC320 work on our borad and get kernel error 

[ 15.549455] GC320 IRQ: 331
[ 15.549464] Galcore version 5.0.11.33433
[ 15.569872] GPU[0](ChipModel=0x320 ChipRevision=0x5301):
[ 15.569884] **************************
[ 15.569890] *** MMU ERROR DUMP ***
[ 15.569895] **************************
[ 15.569904] MMU status = 0x00000000
[ 15.569914] GPU[0](ChipModel=0x320 ChipRevision=0x5301):
[ 15.570632] **************************
[ 15.570640] *** GPU STATE DUMP ***
[ 15.570647] **************************
[ 15.570655] axi = 0x00000001
[ 15.570661] idle = 0x7FFFFFFE
[ 15.570683] FE not idle
[ 15.570691] DMA appears to be stuck at this address:
[ 15.570703] 0x7D4C6010
[ 15.570712] dmaLow = 0x00000000
[ 15.570721] dmaHigh = 0x00000000
[ 15.570726] dmaState = 0x00000000
[ 15.570733] command state = 0 (PAR_IDLE_ST)
[ 15.570740] command DMA state = 0 (CMD_IDLE_ST)
[ 15.570745] command fetch state = 0 (FET_IDLE_ST)
[ 15.570752] DMA request state = 0 (REQ_IDLE_ST)
[ 15.570995] cal state = 0 (CAL_IDLE_ST)
[ 15.571003] VE request state = 0 (VER_IDLE_ST)

I use 

insmod galcore.ko baseAddress=0x80000000 physSize=0x60000000 powerManagement=0

while I check code from

and 

seems bb2d hardware irq route to vivante 3D core soft irq service 

how can I get the bb2d driver run under kernel 4.4?

  • Hi Huang,

    Could you share the details on which SDK version are you running this against?

    Thanks,
    Gowtham
  • Hello Gowtham

    gc320 driver from  git://git.ti.com/graphics/ti-gc320-driver.git  0d62f3a5ab55d3fb476f0107b693379838111ee6

    linux kernel from git://git.omapzoom.org/kernel/omap.git f9b63bd961504c1f6aa7f8708606cdf44dd9d820 

    and our chip information 

    # omapconf read 0x4AE0C20C

    69C70304

    Thanks

  • Hello Gowtham

    I reflash with evm board and with ti original 03_04_00_03 image ,some confuse message

    #cat /proc/interrupts shows
    320: 0 0 CBAR 41 Level rproc-wdt
    323: 0 0 CBAR 339 Level rproc-wdt
    328: 8 0 CBAR 108 Level omap_dmm_irq_handler
    329: 8 0 CBAR 16 Level SGX ISR
    331: 564 0 CBAR 51 Level 48070000.i2c
    332: 59 0 CBAR 52 Level 48072000.i2c
    333: 0 0 CBAR 56 Level 48060000.i2c
    334: 4 0 CBAR 57 Level 4807a000.i2c
    but
    #dmesg shows
    [ 79.823817] GC320 IRQ: 330
    [ 79.826759] Galcore version 5.0.11.33433

    no interrupt register into linux kernel ?Does this normal for ti ?

    Thanks and regards
  • Hello Gowtham

    I recheck it on original bsp filesystem and
    insmod with powerManagement=1
    this will cause the same driver error just with us

    What we want to is disable Powermanager of GC320 , Because from previous experience , in very heavy cpu load and long time test , this will cause kernel crash

    so , Will you try to disable the powermanager in gc320 from kernel driver ?
    or this is just necessory because of ti hardware limit ?

    Thanks and regards
  • Hi Huang,

    > I recheck it on original bsp filesystem and insmod with powerManagement=1 this
    will cause the same driver error just with us

    I didn't follow here. Do you still see error with using the `powerManagement=1`
    option. If so is it the same error as in your first post.

    > What we want to is disable Powermanager of GC320 , Because from previous
    > experience , in very heavy cpu load and long time test , this will cause
    > kernel crash

    Could you share any details on the usecase where it caused issues for you. Note
    that the power manager is on by default and need to check on the disabling it,
    disabling powermanager configuration is not validated on our SDK at this point.

    On a seperate note, how much RAM do you have in your platform configuration. If
    it is > 1.5GB, use `0x80000000` for `physSize` instead of `0x60000000`.

    Thanks,
    Gowtham
  • Hello Gowtham
    Thanks for you reply
    From previous experience , GPU driver will cause kernel crash when enable Power managerment ,In some testcase of 500 hours unit test
    And ,Yes ,we use 2GB ,
    For current now ,we get this work , with 0x80000000 and powermanager on
    just be curious , will you debug the driver with powermanager off ?
    opp might be not needed for our info+ cluster system
    Thanks