On 2019/3/2 0:43, Christoph Muellner wrote:
When using direct commands (DCMDs) on an RK3399, we get spurious CQE completion interrupts for the DCMD transaction slot (#31):
I didn't see it. Do you try any newer code, for instance, linux-next?
[ 931.196520] ------------[ cut here ]------------ [ 931.201702] mmc1: cqhci: spurious TCN for tag 31 [ 931.206906] WARNING: CPU: 0 PID: 1433 at /usr/src/kernel/drivers/mmc/host/cqhci.c:725 cqhci_irq+0x2e4/0x490 [ 931.206909] Modules linked in: [ 931.206918] CPU: 0 PID: 1433 Comm: irq/29-mmc1 Not tainted 4.19.8-rt6-funkadelic #1 [ 931.206920] Hardware name: Theobroma Systems RK3399-Q7 SoM (DT) [ 931.206924] pstate: 40000005 (nZcv daif -PAN -UAO) [ 931.206927] pc : cqhci_irq+0x2e4/0x490 [ 931.206931] lr : cqhci_irq+0x2e4/0x490 [ 931.206933] sp : ffff00000e54bc80 [ 931.206934] x29: ffff00000e54bc80 x28: 0000000000000000 [ 931.206939] x27: 0000000000000001 x26: ffff000008f217e8 [ 931.206944] x25: ffff8000f02ef030 x24: ffff0000091417b0 [ 931.206948] x23: ffff0000090aa000 x22: ffff8000f008b000 [ 931.206953] x21: 0000000000000002 x20: 000000000000001f [ 931.206957] x19: ffff8000f02ef018 x18: ffffffffffffffff [ 931.206961] x17: 0000000000000000 x16: 0000000000000000 [ 931.206966] x15: ffff0000090aa6c8 x14: 0720072007200720 [ 931.206970] x13: 0720072007200720 x12: 0720072007200720 [ 931.206975] x11: 0720072007200720 x10: 0720072007200720 [ 931.206980] x9 : 0720072007200720 x8 : 0720072007200720 [ 931.206984] x7 : 0720073107330720 x6 : 00000000000005a0 [ 931.206988] x5 : ffff00000860d4b0 x4 : 0000000000000000 [ 931.206993] x3 : 0000000000000001 x2 : 0000000000000001 [ 931.206997] x1 : 1bde3a91b0d4d900 x0 : 0000000000000000 [ 931.207001] Call trace: [ 931.207005] cqhci_irq+0x2e4/0x490 [ 931.207009] sdhci_arasan_cqhci_irq+0x5c/0x90 [ 931.207013] sdhci_irq+0x98/0x930 [ 931.207019] irq_forced_thread_fn+0x2c/0xa0 [ 931.207023] irq_thread+0x114/0x1c0 [ 931.207027] kthread+0x128/0x130 [ 931.207032] ret_from_fork+0x10/0x20 [ 931.207035] ---[ end trace 0000000000000002 ]--- The driver shows this message only for the first spurious interrupt by using WARN_ONCE(). Changing this to WARN() shows, that this is happening quite frequently (up to once a second). Since the eMMC 5.1 specification, where CQE and CQHCI are specified, does not mention that spurious TCN interrupts for DCMDs can be simply ignored, we must assume that using this feature is not working reliably. The current implementation uses DCMD for REQ_OP_FLUSH only, and I could not see any performance/power impact when disabling this optional feature for RK3399. Therefore this patch disables DCMDs for RK3399.
We need to sort out the problem, and see if it could be solved, or we just simply remove MMC_CAP2_CQE_DCMD it from sdhci-of-arasan
Signed-off-by: Christoph Muellner <christoph.muellner@xxxxxxxxxxxxxxxxxxxxx> Signed-off-by: Philipp Tomsich <philipp.tomsich@xxxxxxxxxxxxxxxxxxxxx> --- arch/arm64/boot/dts/rockchip/rk3399.dtsi | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/arm64/boot/dts/rockchip/rk3399.dtsi b/arch/arm64/boot/dts/rockchip/rk3399.dtsi index 6cc1c9fa4ea6..1bbf0da4e01d 100644 --- a/arch/arm64/boot/dts/rockchip/rk3399.dtsi +++ b/arch/arm64/boot/dts/rockchip/rk3399.dtsi @@ -333,6 +333,7 @@ phys = <&emmc_phy>; phy-names = "phy_arasan"; power-domains = <&power RK3399_PD_EMMC>; + disable-cqe-dcmd; status = "disabled"; };