On Sat, Oct 05, 2019 at 12:45:29AM +0200, Christian Zigotzky wrote: > Hello, > > The onboard SD card of the Cyrus+ PowerPC board (A-EON AmigaOne X5000) [1] > doesn't work anymore after the 'mmc-v5.4-2' updates [2]. Hi Christian, Please can you help us understand what is going on with your setup. Obviously, your board wants the SDHCI controller to be DMA coherent, but I guess that DT on your platform does not specify the "dma-coherent" property. Can you confirm that? If that's the case, the question is what we do about that - do we add it (and wait for further reports?) The only alternative I can see is that we split ARM usage from PowerPC usage of this driver; it seems PowerPC can get away with being more lenient wrt DT DMA coherency correctness than the ARM architecture. Thanks. > > Error messages: > > [ 12.118148] mmc0: SDHCI controller on ffe114000.sdhc [ffe114000.sdhc] > using ADMA > [ 12.232869] mmc0: ADMA error: 0x02000000 > [ 12.237977] mmc0: sdhci: ============ SDHCI REGISTER DUMP =========== > [ 12.243137] mmc0: sdhci: Sys addr: 0x00000000 | Version: 0x00001301 > [ 12.248307] mmc0: sdhci: Blk size: 0x00000008 | Blk cnt: 0x00000001 > [ 12.253448] mmc0: sdhci: Argument: 0x00000000 | Trn mode: 0x00000013 > [ 12.258530] mmc0: sdhci: Present: 0x01fd020a | Host ctl: 0x00000030 > [ 12.263572] mmc0: sdhci: Power: 0x00000002 | Blk gap: 0x00000000 > [ 12.268574] mmc0: sdhci: Wake-up: 0x00000000 | Clock: 0x000020f8 > [ 12.273535] mmc0: sdhci: Timeout: 0x00000003 | Int stat: 0x00000001 > [ 12.278461] mmc0: sdhci: Int enab: 0x037f008f | Sig enab: 0x037f008b > [ 12.283339] mmc0: sdhci: ACmd stat: 0x00000000 | Slot int: 0x00001301 > [ 12.288198] mmc0: sdhci: Caps: 0x05fa0000 | Caps_1: 0x00000000 > [ 12.293032] mmc0: sdhci: Cmd: 0x0000333a | Max curr: 0x00000000 > [ 12.297861] mmc0: sdhci: Resp[0]: 0x00000920 | Resp[1]: 0x001dd533 > [ 12.302615] mmc0: sdhci: Resp[2]: 0x325b5900 | Resp[3]: 0x3f400e00 > [ 12.307312] mmc0: sdhci: Host ctl2: 0x00000000 > [ 12.311940] mmc0: sdhci: ADMA Err: 0x00000009 | ADMA Ptr: 0x7e04f208 > [ 12.316573] mmc0: sdhci: ============================================ > [ 12.321204] mmc0: sdhci: 7e04f200: DMA 0xfc000000, LEN 0x0008, Attr=0x23 > [ 12.326022] mmc0: error -5 whilst initialising SD card > [ 12.507466] mmc0: invalid bus width > [ 12.512154] mmc0: error -22 whilst initialising SD card > [ 12.683457] mmc0: invalid bus width > [ 12.688968] mmc0: error -22 whilst initialising SD card > [ 12.774281] mmc0: invalid bus width > [ 12.779910] mmc0: error -22 whilst initialising SD card > > ------------------------- > > I have found the issue in the DMA changes of the 'mmc-v5.4-2' updates. It's > the commit "121bd08b029e03404c451bb237729cdff76eafed (mmc: sdhci-of-esdhc: > set DMA snooping based on DMA coherence)" [3]. > > Please find attached a patch. This patch solves the issue. > > dmesg output after compiling with the attached patch: > > [ 12.117149] mmc0: SDHCI controller on ffe114000.sdhc [ffe114000.sdhc] > using ADMA > [ 12.208646] mmc0: new high speed SDHC card at address 59b4 > [ 12.230808] mmcblk0: mmc0:59b4 USD 3.73 GiB > > ------------------------- > > Please check the commit "121bd08b029e03404c451bb237729cdff76eafed" [3]. > > Thanks, > Christian > > > [1] https://www.amigaos.net/hardware/133/amigaone-x5000 and > http://wiki.amiga.org/index.php?title=X5000 > [2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?h=v5.4-rc1&id=c710364f78afdef8c2ed07556d0743c5a30ed429 > [3] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=121bd08b029e03404c451bb237729cdff76eafed > > diff -rupN a/drivers/mmc/host/sdhci-of-esdhc.c b/drivers/mmc/host/sdhci-of-esdhc.c > --- a/drivers/mmc/host/sdhci-of-esdhc.c 2019-10-04 12:40:23.736729250 +0200 > +++ b/drivers/mmc/host/sdhci-of-esdhc.c 2019-10-04 12:40:46.476869118 +0200 > @@ -495,12 +495,7 @@ static int esdhc_of_enable_dma(struct sd > dma_set_mask_and_coherent(dev, DMA_BIT_MASK(40)); > > value = sdhci_readl(host, ESDHC_DMA_SYSCTL); > - > - if (of_dma_is_coherent(dev->of_node)) > - value |= ESDHC_DMA_SNOOP; > - else > - value &= ~ESDHC_DMA_SNOOP; > - > + value |= ESDHC_DMA_SNOOP; > sdhci_writel(host, value, ESDHC_DMA_SYSCTL); > return 0; > } > -- RMK's Patch system: https://www.armlinux.org.uk/developer/patches/ FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up According to speedtest.net: 11.9Mbps down 500kbps up
