On Tue, Oct 04, 2022 at 09:28:16AM -0400, Eric Chanudet wrote: > On Mon, Oct 03, 2022 at 01:31:52PM -0400, Brian Masney wrote: > > Just for documentation purposes, to get linux-next-20220930 booting on > > the QDrive3 with the upstream arm64 defconfig I had to apply the > > following patches: > > > > - arm64: dts: qcom: sc8280xp: fix UFS PHY serdes size > > https://lore.kernel.org/linux-arm-msm/20220915141601.18435-1-johan+linaro@xxxxxxxxxx/ > > > > Without this, the phy fails to probe due to the following error: > > > > qcom-qmp-ufs-phy 1d87000.phy: can't request region for resource [mem 0x01d87400-0x01d87507] > > qcom-qmp-ufs-phy 1d87000.phy: failed to create lane0 phy, -16 > > qcom-qmp-ufs-phy: probe of 1d87000.phy failed with error -16 > > > > - This hack patch is still needed: > > disable has_address_auth_metacap and has_generic_auth > > https://github.com/andersson/kernel/commit/d46a4d05d5a17ff4447af08471edd78e194d48e5 > > > > Without this, the boot hangs at: > > > > rcu: srcu_init: Setting srcu_struct sizes based on contention. > > arch_timer: cp15 and mmio timer(s) running at 19.20MHz (virt/virt). > > clocksource: arch_sys_counter: mask: 0xffffffffffffff max_cycles: 0x46d987e47, max_idle_ns: 440795202767 ns > > sched_clock: 56 bits at 19MHz, resolution 52ns, wraps every 4398046511078ns > > > > - My UFS clock patch is still needed: > > arm64: dts: qcom: sc8280xp: correct ref_aux clock for ufs_mem_phy > > https://lore.kernel.org/lkml/20220830180120.2082734-1-bmasney@xxxxxxxxxx/T/#u > > > > - I didn't use an initrd for testing so I had to change the options > > CONFIG_SCSI_UFS_QCOM and CONFIG_PHY_QCOM_QMP from =m to =y. > > I followed the instructions above and linux-next-20220930 booted on the > QDrive3 to a prompt. It then hanged after a couple minutes and rebooted > in Sahara mode: > B - 1662280 - Sahara Init > B - 1665422 - Sahara Open > > There seems to be no trace from the kernel, this happened consistently > over 3 boots. > > I asked Brian, he mentioned he only booted to prompt so that may have > happened unbeknownst to him as well. I took the upstream defconfig for a spin and see similar. My T-B was for this defconfig (downstream one + make olddefconfig) which works ok for me: https://gitlab.com/ahalaney/linux/-/commit/40f1b241117716ef4b0e27cf50054e749b8ff608 Thanks, Andrew > > -- > Eric Chanudet >
