1553416172@qq.com Posted January 9, 2019 Share Posted January 9, 2019 (edited) I have met an ethernet internal error. error log [ 2.003302] Internal error: Oops: 96000004 [#1] SMP [ 2.008184] Modules linked in: [ 2.011243] CPU: 1 PID: 33 Comm: kworker/1:1 Not tainted 4.19.12-sunxi64 #5.68 [ 2.018458] Hardware name: Orange Pi One+ (DT) [ 2.022909] Workqueue: events deferred_probe_work_func [ 2.028045] pstate: 80000005 (Nzcv daif -PAN -UAO) [ 2.032837] pc : sun8i_dwmac_probe+0x16c/0x550 [ 2.037277] lr : sun8i_dwmac_probe+0x21c/0x550 [ 2.041715] sp : ffff000009273b00 [ 2.045027] x29: ffff000009273b00 x28: 0000000000000000 [ 2.050337] x27: ffff000008d9eb30 x26: ffff000008c023a8 [ 2.055647] x25: ffff800077fe2308 x24: ffff000008c58000 [ 2.060957] x23: ffff800075164810 x22: ffff800075164800 [ 2.066267] x21: ffff800009a15c18 x20: ffff800075059c98 [ 2.071577] x19: ffff000008d88000 x18: ffff000008d90000 [ 2.076887] x17: 0000000000000001 x16: 0000000000000019 [ 2.082197] x15: ffff000008daa000 x14: 2e726f74616c7567 [ 2.087507] x13: ffff000008c96000 x12: 0000000000000030 [ 2.092816] x11: 0101010101010101 x10: 7f7f7f7f7f7f7f7f [ 2.098126] x9 : feff727264626260 x8 : ffff0000092737c8 [ 2.103435] x7 : 0000000000000000 x6 : 0000000000000003 [ 2.108745] x5 : 0000000000000002 x4 : 0000000000000003 [ 2.114055] x3 : 0000000000000000 x2 : ffff000009273b70 [ 2.119365] x1 : ffff800009a16000 x0 : ffff800075164810 [ 2.124679] Process kworker/1:1 (pid: 33, stack limit = 0x(____ptrval____)) [ 2.131635] Call trace: [ 2.134084] sun8i_dwmac_probe+0x16c/0x550 [ 2.138182] platform_drv_probe+0x50/0xa0 [ 2.142193] really_probe+0x1c4/0x298 [ 2.145857] driver_probe_device+0x58/0x100 [ 2.150039] __device_attach_driver+0x98/0xf0 [ 2.154397] bus_for_each_drv+0x64/0xc8 [ 2.158234] __device_attach+0xd8/0x130 [ 2.162070] device_initial_probe+0x10/0x18 [ 2.166254] bus_probe_device+0x90/0x98 [ 2.170089] deferred_probe_work_func+0x74/0xb0 [ 2.174621] process_one_work+0x1e4/0x360 [ 2.178631] worker_thread+0x268/0x4b0 [ 2.182381] kthread+0x128/0x130 [ 2.185611] ret_from_fork+0x10/0x1c [ 2.189188] Code: f9401283 9101c3a2 aa1703e0 f9400463 (a9401464) [ 2.195279] ---[ end trace 2d0d6db1bbcb41cc ]--- I try to solve this error, but I can not modify source code. running compile.sh scripts makes patch cover my work. The only way I know is to modify patch which makes debug process complicated and inconvinent. Furthermore, every time I running compiles.sh, it seems that most part of source code are downloaded from scratch. That waste a lot of time. So, I wander if there is a way just to modify source code and compile. With banning patch cover and redownload source code Or, is another way to fixed this error. A guy met this error,link:https://forum.armbian.com/topic/9131-h6-after-apt-upgrade/. I can not find a way to switch kernel version to 4.18.0-rc7 Edited January 9, 2019 by 1553416172@qq.com Link to comment Share on other sites More sharing options...
guidol Posted January 9, 2019 Share Posted January 9, 2019 4 hours ago, 1553416172@qq.com said: I try to solve this error, but I can not modify source code. running compile.sh scripts makes patch cover my work. The only way I know is to modify patch which makes debug process complicated and inconvinent. with "modify patch" do you mean the patch-option inside ./build/config-default.conf ? CREATE_PATCHES="yes" # wait that you make changes to uboot and kernel source and creates patches Maybe you have less download when setting CLEAN_LEVEL to another value? CLEAN_LEVEL="make,debs,oldcache" # comma-separated list of clean targets: "make" = make clean for selected kernel and u-boot, # "debs" = delete packages in "./output/debs" for current branch and family, # "alldebs" = delete all packages in "./output/debs", "images" = delete "./output/images", # "cache" = delete "./output/cache", "sources" = delete "./sources" # "oldcache" = remove old cached rootfs except for the newest 6 files Link to comment Share on other sites More sharing options...
Werner Posted January 9, 2019 Share Posted January 9, 2019 After you run compile.sh the first time a userpatches directory is created where you can place own patches. Until now the only kernel known to work is 4.18.0-rc7. Quote A guy met this error,link:https://forum.armbian.com/topic/9131-h6-after-apt-upgrade/. I can not find a way to switch kernel version to 4.18.0-rc7 I do not know if pure Debian packages with this specific kernel image still existing somewhere, but AFAIK are the latest precompiled images for OPi 1+ and Lite2 are still shipped with this kernel. So if you have one of these boards you may start over and freeze the kernel from being upgraded using armbian-config Link to comment Share on other sites More sharing options...
martinayotte Posted January 9, 2019 Share Posted January 9, 2019 12 hours ago, 1553416172@qq.com said: I have met an ethernet internal error. For this crash, you don't need to recompile kernel to get rid of it ! Simply edit DT to "disable" the "ethernet" ... Link to comment Share on other sites More sharing options...
1553416172@qq.com Posted January 10, 2019 Author Share Posted January 10, 2019 19 hours ago, guidol said: with "modify patch" do you mean the patch-option inside ./build/config-default.conf ? CREATE_PATCHES="yes" # wait that you make changes to uboot and kernel source and creates patches Maybe you have less download when setting CLEAN_LEVEL to another value? CLEAN_LEVEL="make,debs,oldcache" # comma-separated list of clean targets: "make" = make clean for selected kernel and u-boot, # "debs" = delete packages in "./output/debs" for current branch and family, # "alldebs" = delete all packages in "./output/debs", "images" = delete "./output/images", # "cache" = delete "./output/cache", "sources" = delete "./sources" # "oldcache" = remove old cached rootfs except for the newest 6 files modified config-default.conf, changed master to sunxi-4.18 #LIB_TAG="master" # change to "branchname" to use any branch currently available. LIB_TAG="sunxi-4.18" # change to "branchname" to use any branch currently available. and compiled, found out that cache/sources/linux-mainline/orange-pi-4.19/ still being used to compiled. Does it mean kernel version is 4.19 ? Link to comment Share on other sites More sharing options...
1553416172@qq.com Posted January 10, 2019 Author Share Posted January 10, 2019 11 hours ago, martinayotte said: For this crash, you don't need to recompile kernel to get rid of it ! Simply edit DT to "disable" the "ethernet" ... Thanks, good to know how to disable ethernet, but have to use ethernet on board. So have to find a way to fixed this bug Link to comment Share on other sites More sharing options...
1553416172@qq.com Posted January 10, 2019 Author Share Posted January 10, 2019 18 hours ago, Werner said: After you run compile.sh the first time a userpatches directory is created where you can place own patches. Until now the only kernel known to work is 4.18.0-rc7. I do not know if pure Debian packages with this specific kernel image still existing somewhere, but AFAIK are the latest precompiled images for OPi 1+ and Lite2 are still shipped with this kernel. So if you have one of these boards you may start over and freeze the kernel from being upgraded using armbian-config modify code and then using git to generate patches, copy or move patches to userpatches directory. Is it a correct manner for employing userpatched? and still confused with armbian-config, what is it? And how to use. Does armbian-config point to config-default.conf Link to comment Share on other sites More sharing options...
Werner Posted January 10, 2019 Share Posted January 10, 2019 Yes, I think this is the way to go. armbian-config has nothing to do with the build system. It is a independent tool to set various system values, switch/freeze kernel version, and so on within a running Armbian system. Sorry for confusion. Link to comment Share on other sites More sharing options...
martinayotte Posted January 10, 2019 Share Posted January 10, 2019 13 hours ago, 1553416172@qq.com said: good to know how to disable ethernet, but have to use ethernet on board. So have to find a way to fixed this bug We, Armbian Devs, didn't found a way to fix that issue yet, so you will need another way to get ethernet, such as USB adaptor ... 1 Link to comment Share on other sites More sharing options...
Tido Posted January 14, 2019 Share Posted January 14, 2019 @1553416172@qq.com Planned for Kernel 5.0 H6 Ethernet DE3/HDMI support USB 2.0 http://linux-sunxi.org/Linux_mainlining_effort#Planned_for_5.0 Link to comment Share on other sites More sharing options...
martinayotte Posted January 14, 2019 Share Posted January 14, 2019 1 hour ago, Tido said: Planned for Kernel 5.0 As soon as those patches will come in, we will probably try to backport them into 4.20.y ... Link to comment Share on other sites More sharing options...
Werner Posted January 21, 2019 Share Posted January 21, 2019 @martinayotte May you take a look here: https://github.com/EvilOlaf/build/commit/eb4a36885bc63199bbc5aadb2bd220c560af0109#commitcomment-32007637 Is this something known or worth trying? Link to comment Share on other sites More sharing options...
martinayotte Posted January 21, 2019 Share Posted January 21, 2019 6 hours ago, Werner said: Is this something known or worth trying? I don't know, but this leave me in big doubts, since this 0x03000030 doesn't point to any kind of registers in the Allwinner datasheets or specs Quote emac: ethernet@5020000 { compatible = "allwinner,sun50i-a64-emac"; syscon = <&syscon>; reg = <0x05020000 0x10000>, <0x0 0x03000030 0x4>; interrupts = <GIC_SPI 12 IRQ_TYPE_LEVEL_HIGH>; EDIT : It seems that only "compatible" change is required, although this is pretty much a hack, since it doesn't use H6 compatible that @Icenowy delivered ... 1 Link to comment Share on other sites More sharing options...
Icenowy Posted January 22, 2019 Share Posted January 22, 2019 @martinayotte I remember in mainline Linuc I killed sun50i-h6-emac in driver, and it only exists as a fallback compatible string. 1 Link to comment Share on other sites More sharing options...
martinayotte Posted January 22, 2019 Share Posted January 22, 2019 14 hours ago, Icenowy said: I killed sun50i-h6-emac So, do you mean that the current solution is simply to use "allwinner,sun50i-a64-emac" for now instead of "allwinner,sun50i-h6-emac" ? Link to comment Share on other sites More sharing options...
1553416172@qq.com Posted January 29, 2019 Author Share Posted January 29, 2019 On 1/22/2019 at 5:06 AM, martinayotte said: I don't know, but this leave me in big doubts, since this 0x03000030 doesn't point to any kind of registers in the Allwinner datasheets or specs EDIT : It seems that only "compatible" change is required, although this is pretty much a hack, since it doesn't use H6 compatible that @Icenowy delivered ... 0x03000030 points to ephy register which H6 specification does not mention. Refer to kernel/drivers/net/phy/sunxi-ephy.c kernel/drivers/mfd/acx00-core.c and kernel/arch/arm64/boot/dts/sun50iw6p1.dtsi, if you can find linux 3.10.65 from allwinner. Ethernet don't work with linux 4.19 because of missing phy driver. My boss make me to give it up, cause wasting a lot of time. So, I have no time to transplant phy driver to linux 4.19. You can refer to H6 spercs 3.20 ATE(audio codec, tve, ephy) controller、8.8 emac and 8.13 ephy setions from Allwinner_H6 V200_User_Manual_V1.1.pdf Looking forward to you work Allwinner_H6 V200_User_Manual_V1.1.pdf @martinayotte 1 Link to comment Share on other sites More sharing options...
belfastraven Posted January 30, 2019 Share Posted January 30, 2019 If it helps anyone, I've been running succesfully with gigabit ethernet on eth0 using the dts and h6 dtsi from Icenowy's aosc-sunxi64-4.19-malimidgard-hack-2 branch and the armbian dev kernel . I force a stable mac_address by adding allow-hotplug eth0 iface eth0 inet dhcp hwaddress ether xx:xx:xx:xx:xx:xx (replace the xx's of course with a good address) in the /etc/network/interfaces file. I haven't yet been able to get spi-jedec-nor flashed to enable sdi boot, I get an error indicating that there is a mismatch between what is specified in the DTS vs the overlay, but I am booting off of an sd card with the file system on a usb connected ssd. I'm going to try to build with the mainline 5 kernel soon, to see If I can pick up the new graphics drivers... I believe version B of the device is supposed to be out soon... 1 Link to comment Share on other sites More sharing options...
Werner Posted February 2, 2019 Share Posted February 2, 2019 On 1/30/2019 at 3:07 AM, belfastraven said: If it helps anyone, I've been running succesfully with gigabit ethernet on eth0 using the dts and h6 dtsi from Icenowy's aosc-sunxi64-4.19-malimidgard-hack-2 branch and the armbian dev kernel . I force a stable mac_address by adding allow-hotplug eth0 iface eth0 inet dhcp hwaddress ether xx:xx:xx:xx:xx:xx (replace the xx's of course with a good address) in the /etc/network/interfaces file. I haven't yet been able to get spi-jedec-nor flashed to enable sdi boot, I get an error indicating that there is a mismatch between what is specified in the DTS vs the overlay, but I am booting off of an sd card with the file system on a usb connected ssd. I'm going to try to build with the mainline 5 kernel soon, to see If I can pick up the new graphics drivers... I believe version B of the device is supposed to be out soon... May you share your working configuration as PR to armbian-build on Github? H6 is still WIP and I guess the cutback that comes with it (hwaddress) is justifiable ... and it is still development, right? Link to comment Share on other sites More sharing options...
Recommended Posts