On Thu, Jul 28, 2022 at 10:12:11AM -0700, Tadeusz Struk wrote:

On 7/28/22 07:45, Greg Kroah-Hartman wrote:

...

On Thu, Jul 28, 2022 at 10:23:18AM -0400, Tim Lewis wrote:

...

Also trying to mount btrfs after boot:

[ 2075.535319] Btrfs loaded, crc32c=crc32c-generic [ 2075.537092] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000068 [ 2075.540237] Mem abort info: [ 2075.542974] ESR = 0x96000004 [ 2075.546591] EC = 0x25: DABT (current EL), IL = 32 bits [ 2075.551332] SET = 0, FnV = 0 [ 2075.554309] EA = 0, S1PTW = 0 [ 2075.558240] Data abort info: [ 2075.560259] ISV = 0, ISS = 0x00000004 [ 2075.564102] CM = 0, WnR = 0 [ 2075.566962] user pgtable: 4k pages, 48-bit VAs, pgdp=000000001e0f3000 [ 2075.573388] [0000000000000068] pgd=0000000000000000, p4d=0000000000000000 [ 2075.580098] Internal error: Oops: 96000004 [#1] PREEMPT SMP [ 2075.585581] Modules linked in: btrfs blake2b_generic zstd_compress raid0 raid1 dm_raid raid456 async_raid6_recov async_memcpy async_pq md_mod raid6_pq dm_integrity async_xor async_tx xor xor_neon xfs bridge stp llc ip6t_REJECT ip6table_filter ip6table_mangle ip6_tables xt_set ipt_REJECT xt_conntrack iptable_filter xt_MASQUERADE iptable_nat xt_CHECKSUM xt_tcpudp iptable_mangle ip_set_hash_ip ip_set_hash_net ip_set snd_soc_hdmi_codec dw_hdmi_i2s_audio dw_hdmi_cec dwmac_generic meson_gxl realtek dwmac_meson8b meson_dw_hdmi stmmac_platform axg_audio rc_odroid dw_hdmi snd_soc_meson_g12a_tohdmitx stmmac sclk_div snd_soc_meson_axg_sound_card panfrost meson_rng meson_drm meson_ir snd_soc_meson_axg_tdm_interface crct10dif_ce rc_core snd_soc_meson_axg_tdmout reset_meson_audio_arb cec snd_soc_meson_codec_glue snd_soc_meson_axg_frddr rtc_meson_vrtc rng_core gpu_sched clk_phase meson_canvas snd_soc_meson_axg_fifo drm_kms_helper snd_soc_meson_card_utils pcs_xpcs nvmem_meson_efuse [ 2075.585716] snd_soc_meson_axg_tdm_formatter display_connector nft_reject_inet nf_reject_ipv6 nft_reject drm ip_tables x_tables ipv6 [ 2075.682357] CPU: 0 PID: 5100 Comm: systemd-udevd Not tainted 5.10.134-rc1-dirty #1 [ 2075.689858] Hardware name: Hardkernel ODROID-C4 (DT) [ 2075.694776] pstate: 40400009 (nZcv daif +PAN -UAO -TCO BTYPE=--) [ 2075.700732] pc : bio_alloc_bioset+0x20/0x270 [ 2075.704980] lr : btrfs_alloc_device+0x64/0x190 [btrfs] [ 2075.710040] sp : ffff80001500bc60 [ 2075.713317] x29: ffff80001500bc60 x28: ffff8000095a1000 [ 2075.718578] x27: 0000000000000000 x26: ffff00001cf8ec78 [ 2075.723839] x25: 0000000000032e45 x24: ffff000000ca6580 [ 2075.729101] x23: 0000000000000000 x22: 0000000000000000 [ 2075.734362] x21: ffff000000ca6580 x20: 0000000000000000 [ 2075.739623] x19: ffff00001cf8ee00 x18: 0000000000000000 [ 2075.744884] x17: 0000000000000000 x16: 0000000000000000 [ 2075.750146] x15: 0000000000000000 x14: 0000000000000000 [ 2075.755407] x13: 0000000000000002 x12: 0000000000000000 [ 2075.760668] x11: 0000000000000001 x10: 0000000000000a30 [ 2075.765929] x9 : 0000000000000000 x8 : ffff00001cf8f000 [ 2075.771191] x7 : 0000000000000000 x6 : 000000000000003f [ 2075.776452] x5 : 0000000000000040 x4 : 0000000000000000 [ 2075.781713] x3 : 0000000000000004 x2 : 0000000000000000 [ 2075.786974] x1 : 0000000000000000 x0 : 0000000000000cc0 [ 2075.792237] Call trace: [ 2075.794657] bio_alloc_bioset+0x20/0x270 [ 2075.798560] btrfs_alloc_device+0x64/0x190 [btrfs] [ 2075.803299] device_list_add+0x418/0x6b8 [btrfs] [ 2075.807870] btrfs_scan_one_device+0xd8/0x208 [btrfs] [ 2075.812876] btrfs_control_ioctl+0xdc/0x188 [btrfs] [ 2075.817686] __arm64_sys_ioctl+0xa8/0xf0 [ 2075.821565] el0_svc_common.constprop.0+0x78/0x1c8 [ 2075.826306] do_el0_svc+0x24/0x90 [ 2075.829585] el0_svc+0x14/0x20 [ 2075.832602] el0_sync_handler+0xb0/0xb8 [ 2075.836397] el0_sync+0x180/0x1c0 [ 2075.839677] Code: aa0203f4 a90363f7 2a0103f7 d5384118 (f9403441) [ 2075.845714] ---[ end trace fd4b78de8314a3ec ]--- [10758.363281] Process accounting resumed

There was a change to bio_alloc_bioset:

https://lore.kernel.org/all/20220727161012.364032637@linuxfoundation.org/

Thanks, I've now dropped that one and some others and will push out a -rc2 in a few minutes...

Sorry about the hassle. It was only tested on ext4 fs.

Tim, The fix consisted of two patches. I'm just curies if it helps if you apply this on top:

https://lore.kernel.org/all/20220727161012.411398366@linuxfoundation.org/

That was "on top" when others reported it failing. And there were at least 2 other needed fixes on top of that :(

Please work to make a full set of patches that are tested and I will be glad to reconsider, otherwise this just isn't going to be an issue for 5.10.y that we need to worry about. If you are concerned about this problem, use 5.15.y or newer please.

thanks,

greg k-h