On Tue, Mar 29, 2022 at 01:02:34AM +0100, Mark Brown wrote: > On Sun, Sep 05, 2021 at 05:01:18PM +0100, Steve McIntyre wrote: > I bisected this to > > commit 7a8b64d17e35810dc3176fe61208b45c15d25402 > > of/address: use range parser for of_dma_get_range > > on what appears to have been a DT system with a built in DT from > the firmware, using upstream's defconfig. I dimly recall seeing Oh, and probably worth pointing out that current mainline handles the failures with ATA a lot more gracefully, still failing to bring up ATA but just printing: [ 2.652674] ahci 0000:02:00.0: SSS flag set, parallel bus scan disabled [ 2.659364] ahci 0000:02:00.0: AHCI 0001.0200 32 slots 2 ports 6 Gbps 0x3 impl SATA mode [ 2.667474] ahci 0000:02:00.0: flags: 64bit ncq sntf stag led clo pmp pio slum part ccc sxs [ 2.676825] ahci 0000:02:00.0: failed to start port 0 (errno=-12) [ 2.682955] ahci: probe of 0000:02:00.0 failed with error -12 and MMC still also fails but this time with an identified oops rather than just timeouts: [ 2.987953] ------------[ cut here ]------------ [ 2.994834] usbcore: registered new interface driver usbhid [ 2.998370] f_sdh30 52300000.sdhci: swiotlb addr 0xffffffffffffffff+65536 overflow (mask ffffffff, bus limit 0). [ 2.998395] WARNING: CPU: 2 PID: 8 at kernel/dma/swiotlb.c:740 swiotlb_map+0x1e4/0x1f8 [ 3.003981] usbhid: USB HID core driver [ 3.014149] Modules linked in: [ 3.014159] CPU: 2 PID: 8 Comm: kworker/u48:0 Not tainted 5.17.0-11137-gb953c6822375 #1 [ 3.026681] NET: Registered PF_PACKET protocol family [ 3.028964] Hardware name: Socionext SynQuacer E-series DeveloperBox, BIOS build #54 Mar 6 2019 [ 3.028973] Workqueue: events_unbound async_run_entry_fn [ 3.028987] pstate: 60000005 (nZCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--) [ 3.037169] 9pnet: Installing 9P2000 support [ 3.042056] pc : swiotlb_map+0x1e4/0x1f8 [ 3.042068] lr : swiotlb_map+0x1e4/0x1f8 [ 3.042078] sp : ffff80000818bb80 [ 3.050930] Key type dns_resolver registered [ 3.056180] x29: ffff80000818bb80 x28: 0000000000000000 x27: 0000000000000000 [ 3.063504] registered taskstats version 1 [ 3.067423] [ 3.067427] x26: 0000000000010000 x25: 0000000000000080 x24: ffff000801787010 [ 3.067439] x23: 0000000000000000 x22: 0000000000000000 [ 3.071388] Loading compiled-in X.509 certificates [ 3.075292] x21: 0000000000010000 [ 3.075299] x20: ffff8000098c3998 x19: ffff000801787010 [ 3.079127] pstore: Using crash dump compression: deflate [ 3.082890] x18: 0000000000000020 [ 3.082897] x17: 0000000000000000 x16: 206f74206b636162 x15: ffffffffffffffff [ 3.082909] x14: 0000000000000001 [ 3.097298] OF: translation of DMA address(0) to CPU address failed node() [ 3.102762] x13: ffff000f7bf76556 x12: ffff000f7bf76551 [ 3.102773] x11: 0720072007200720 x10: 000000000000000a x9 : ffff80000818bb80 [ 3.102785] x8 : ffff80000a331198 x7 : ffff80000818b980 [ 3.108037] gpio-keys gpio-keys: Invalid size 0x1 for dma-range(s) [ 3.112816] x6 : ffff000f7bf3ef80 [ 3.112823] x5 : 0000000000000000 x4 : 0000000000000000 [ 3.116564] input: gpio-keys as /devices/platform/gpio-keys/input/input0 [ 3.121457] x3 : 00000000ffffffff [ 3.121464] x2 : ffff80000a0125a0 x1 : fbd95b9fae668f00 x0 : 0000000000000000 [ 3.197275] Call trace: [ 3.199721] swiotlb_map+0x1e4/0x1f8 [ 3.203305] dma_map_page_attrs+0xec/0x228 [ 3.207408] sdhci_setup_host+0x8f0/0xd30 [ 3.211430] sdhci_add_host+0x18/0x50 [ 3.215098] sdhci_f_sdh30_probe+0x2a0/0x378 [ 3.219373] platform_probe+0x68/0xd8 [ 3.223043] really_probe+0xb8/0x300 [ 3.226622] __driver_probe_device+0x78/0xe0 [ 3.230896] driver_probe_device+0x3c/0xe8 [ 3.234997] __driver_attach_async_helper+0x2c/0x50 [ 3.239880] async_run_entry_fn+0x34/0xd0 [ 3.243896] process_one_work+0x1ec/0x330 [ 3.247912] worker_thread+0x44/0x420 [ 3.251579] kthread+0x110/0x120 [ 3.254815] ret_from_fork+0x10/0x20 [ 3.258397] ---[ end trace 0000000000000000 ]--- followed by more in sdhci_send_command(). [ 23.661522] WARNING: CPU: 0 PID: 0 at drivers/mmc/host/sdhci.c:1151 sdhci_send_command+0x954/0xde8 I'm guessing these are going to be the same underlying issue with the firmware but manifesting a bit differently, the network device does still complain about the 1 byte dma-range. The system gets to a ramdisk but has no useful I/O.
Attachment:
signature.asc
Description: PGP signature