Bug#750473: u-boot: Please enable Cubietruck support
Hello Vagrant! On Sat, Jun 14, 2014 at 03:02:02PM -0700, Vagrant Cascadian wrote: Perhaps against my better judgement again, I spent a little more time on it, and have packages available to test: http://people.debian.org/~vagrant/u-boot/ http://people.debian.org/~vagrant/u-boot/u-boot_2014.04-3~20140614~1_armhf.deb Hehe, sorry for pushing you into doing that! :) I'm just building my own u-boot package right now, where I've backported the AHCI patches (+ the MMC DMA removal patch). :P I'll test both yours and mine as soon as possible. Sorry for taking up so much of your time. I really appreciate your interest in making Cubietruck support available though! Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
Hello again Vagrant! First the bad news, then the good... Unfortunately your new version does not seem to include AHCI support. On Sat, Jun 14, 2014 at 03:02:02PM -0700, Vagrant Cascadian wrote: [...] I took all patches from v2014.07-rc2 to the HEAD of that branch, and other than boards.cfg, seemed to apply and build fine. Possibly because of a mismerge of boards.cfg? ah@heavy:/tmp/vagrant/u-boot-2014.04$ grep Cubietruck boards.cfg Active arm armv7 sunxi - sunxi Cubietrucksun7i:CUBIETRUCK,SPL,SUNXI_GMAC,RGMII - Active arm armv7 sunxi - sunxi Cubietruck_FEL sun7i:CUBIETRUCK,SPL_FEL,SUNXI_GMAC,RGMII - There's no mention of AHCI or SATAPWR=... in the above lines. A smaller patchset would be preferable, but at least most of the changes are sunxi-specific. They were applied on top of the Cubietruck patches already applied from upstream. [...] I'm attaching two patches which is against http://anonscm.debian.org/gitweb/?p=collab-maint/u-boot.git;a=summary commit 3764e45bbcbc858ecc83ec Updated lintian overrides: which is the patches from the mailing list with minor massage to apply without all the other sunxi boards which apparently got sent for review between what's already merged upstream and the AHCI thread. Please confirm if it works for you, or have suggestions or an alternate patchset to apply. Unfortunately your u-boot does not include the scsi command at all. My version finds my SSD when I do scsi scan. Please tell me if you'd rather see me push straight to collab-maint then to mail you patches Regards, Andreas Henriksson From 9221c88f68905f8f8411d31447019cbe6e73ed13 Mon Sep 17 00:00:00 2001 From: Andreas Henriksson andr...@fatal.se Date: Sun, 15 Jun 2014 08:08:17 +0200 Subject: [PATCH 1/2] Add patches for sunxi AHCI driver (Cubietruck) This is patch 1,2,4,5,6 (skipping 3 - Cubietruck2) from: http://lists.denx.de/pipermail/u-boot/2014-June/181047.html With minor modifications to get it to apply cleanly. --- debian/patches/series | 5 + .../0013-AHCI-Increase-link-timeout-to-200ms.patch | 31 +++ .../0014-board_r-run-scsi-init-on-ARM-too.patch| 52 .../0015-sunxi-add-gpio-driver.patch | 175 ++ ...-use-setbits_le32-to-enable-the-DMA-clock.patch | 24 ++ ...e-sunxi-SATA-driver-using-AHCI-platform-f.patch | 269 + 6 files changed, 556 insertions(+) create mode 100644 debian/patches/sunxi-cubietruck/0013-AHCI-Increase-link-timeout-to-200ms.patch create mode 100644 debian/patches/sunxi-cubietruck/0014-board_r-run-scsi-init-on-ARM-too.patch create mode 100644 debian/patches/sunxi-cubietruck/0015-sunxi-add-gpio-driver.patch create mode 100644 debian/patches/sunxi-cubietruck/0016-sunxi-use-setbits_le32-to-enable-the-DMA-clock.patch create mode 100644 debian/patches/sunxi-cubietruck/0017-ahci-provide-sunxi-SATA-driver-using-AHCI-platform-f.patch diff --git a/debian/patches/series b/debian/patches/series index e8e24dc..2f6beeb 100644 --- a/debian/patches/series +++ b/debian/patches/series @@ -26,5 +26,10 @@ sunxi-cubietruck/0009-net-designware-ensure-device-private-data-is-DMA-ali.patch sunxi-cubietruck/0010-net-designware-ensure-cache-invalidations-are-aligne.patch sunxi-cubietruck/0011-net-designware-reorder-struct-dw_eth_dev-to-pack-mor.patch sunxi-cubietruck/0012-net-designware-Make-DMA-burst-length-configurable-an.patch +sunxi-cubietruck/0013-AHCI-Increase-link-timeout-to-200ms.patch +sunxi-cubietruck/0014-board_r-run-scsi-init-on-ARM-too.patch +sunxi-cubietruck/0015-sunxi-add-gpio-driver.patch +sunxi-cubietruck/0016-sunxi-use-setbits_le32-to-enable-the-DMA-clock.patch +sunxi-cubietruck/0017-ahci-provide-sunxi-SATA-driver-using-AHCI-platform-f.patch am335x-bootscript.diff am335x-try-boot-from-first-partition.diff diff --git a/debian/patches/sunxi-cubietruck/0013-AHCI-Increase-link-timeout-to-200ms.patch b/debian/patches/sunxi-cubietruck/0013-AHCI-Increase-link-timeout-to-200ms.patch new file mode 100644 index 000..b22d52c --- /dev/null +++ b/debian/patches/sunxi-cubietruck/0013-AHCI-Increase-link-timeout-to-200ms.patch @@ -0,0 +1,31 @@ +From: Ian Campbell i...@hellion.org.uk +Date: Thu, 5 Jun 2014 19:00:12 +0100 +Subject: AHCI: Increase link timeout to 200ms + +In 73545f75b66d ahci: wait longer for link I increased the +timeout to 40ms based on the observed behaviour of a WD disk on a +Cubietruck. Since then Karsten Merker and myself have both +observed timeouts with HGST disks (Karsten on Cubietruck, me on +Cubieboard2). Increasing the timeout to ~175ms fixes this, so go +to 200ms for a bit of headroom. + +Signed-off-by: Ian Campbell
Bug#750473: u-boot: Please enable Cubietruck support
Hello again! I briefly met Vagrant on IRC and he figured out that in my massive confusion caused by a flaky usb-mmcreader-dongle, I had managed to miss that there where 2 different versions available at http://people.debian.org/~vagrant/u-boot/ and that I was looking at the old one. Sorry for that! (Atleast I managed to consistently test the old binary and look at the old source...) Anyway, the result with u-boot_2014.04-3~20140614~1_armhf.deb is that it gets stuck at bootup before reaching the prompt. All I can see is: U-Boot SPL 2014.04 (Jun 14 2014 - 19:54:04) Vagrant also said if your patchset works, that's preferable anyways and yes the 2 patches I sent in my previous mail works, so I guess we'll go with those for now then... Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
Control: tags 750473 pending On Sun, Jun 15, 2014 at 10:03:25AM +0200, Andreas Henriksson wrote: A smaller patchset would be preferable, but at least most of the changes are sunxi-specific. They were applied on top of the Cubietruck patches already applied from upstream. ... My version finds my SSD when I do scsi scan. Please tell me if you'd rather see me push straight to collab-maint then to mail you patches A branch in collab-maint would have been great, but for now I've applied and will push them, and hopefully upload soon... Now you're on the hook for keeping them updated until upstream inclusion! :) live well, vagrant signature.asc Description: Digital signature
Bug#750473: u-boot: Please enable Cubietruck support
On Thu, Jun 12, 2014 at 04:09:11PM -0700, Vagrant Cascadian wrote: On Thu, Jun 12, 2014 at 09:39:33PM +0200, Andreas Henriksson wrote: [...] I think Ian Campbell also had a few good points about SATA support so if you haven't already considered it I would recommend looking at http://lists.denx.de/pipermail/u-boot/2014-June/181047.html (Hopefully these will be merged upstream any day now..) Just waiting for upstream inclusion, unless they're simple and non-invasive. IMHO they could not be much more simple an non-invasive, so I'm still hoping you can include them. They simply hook up an existing driver and fixes a couple of timeout settings. (You are also already carrying code which is not simple an un-invasive and have a much longer road towards being included upstream, so I'm having a hard time understanding why adding these would be a problem. Please enlighten me so I can help solve it! Unfortunately getting upstream u-boot developers to review/apply patches quicker then they are already doing is not something I can help with.) If you need to be convinced of the usefulness of the AHCI support in Debians u-boot, please see https://wiki.debian.org/InstallingDebianOn/Allwinner where you can read about the biggest gotchas with installing Debian on Allwinner hardware. That would be u-boot AHCI support and Linux MMC support -- where as MMC support is now solved with 3.15~rc8. Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
On Sat, Jun 14, 2014 at 03:12:39PM +0200, Andreas Henriksson wrote: On Thu, Jun 12, 2014 at 04:09:11PM -0700, Vagrant Cascadian wrote: On Thu, Jun 12, 2014 at 09:39:33PM +0200, Andreas Henriksson wrote: [...] I think Ian Campbell also had a few good points about SATA support so if you haven't already considered it I would recommend looking at http://lists.denx.de/pipermail/u-boot/2014-June/181047.html (Hopefully these will be merged upstream any day now..) Just waiting for upstream inclusion, unless they're simple and non-invasive. IMHO they could not be much more simple an non-invasive, so I'm still hoping you can include them. They simply hook up an existing driver and fixes a couple of timeout settings. The upstreamed patches for Cubietruck I managed to grab and backport trivially, and someone verified that they work, so I will include that much. If someone wants ahci support included, please backport the patch series to the Debian package and the submit patches. My attempts so far have taken longer than I'd like, and even if I got the patches to apply, I don't have the hardware to verify that they work correctly. (You are also already carrying code which is not simple an un-invasive and have a much longer road towards being included upstream, so I'm having a hard time understanding why adding these would be a problem. Please enlighten me so I can help solve it! The difference with the rather large patchset for cubox-i is that someone was willing to provide the patches and commit to keep them working with new upstream versions, and I was personally able to test that it works, as I have the hardware. The fact that it's not upstreamed and doesn't appear to be moving upstream does make me a bit nervous... maybe it was not good judgement on my part. :/ live well, vagrant signature.asc Description: Digital signature
Bug#750473: u-boot: Please enable Cubietruck support
On Sat, Jun 14, 2014 at 11:03:59AM -0700, Vagrant Cascadian wrote: If someone wants ahci support included, please backport the patch series to the Debian package and the submit patches. My attempts so far have taken longer than I'd like, and even if I got the patches to apply, I don't have the hardware to verify that they work correctly. Perhaps against my better judgement again, I spent a little more time on it, and have packages available to test: http://people.debian.org/~vagrant/u-boot/ http://people.debian.org/~vagrant/u-boot/u-boot_2014.04-3~20140614~1_armhf.deb The patches were taken from: git://gitorious.org/ijc/u-boot.git branch: sunxi-gpio-ahci-and-cubieboard2-v2 I took all patches from v2014.07-rc2 to the HEAD of that branch, and other than boards.cfg, seemed to apply and build fine. A smaller patchset would be preferable, but at least most of the changes are sunxi-specific. They were applied on top of the Cubietruck patches already applied from upstream. b/arch/arm/cpu/armv7/sunxi/board.c | 13 b/arch/arm/cpu/armv7/sunxi/clock_sun4i.c |3 b/arch/arm/cpu/armv7/sunxi/u-boot-spl.lds|1 b/arch/arm/include/asm/arch-sunxi/gpio.h |2 b/arch/arm/include/asm/arch-sunxi/i2c.h | 15 b/arch/arm/include/asm/arch-sunxi/timer.h|5 b/board/sunxi/board.c|7 b/board/sunxi/gmac.c | 11 b/common/board_r.c |6 b/drivers/block/ahci.c |2 b/drivers/gpio/Makefile |1 b/drivers/gpio/sunxi_gpio.c | 102 +++ b/drivers/i2c/Makefile |1 b/drivers/i2c/mvtwsi.c | 18 b/drivers/mmc/sunxi_mmc.c| 140 - b/drivers/net/Makefile |2 b/drivers/net/sunxi_emac.c | 521 ++ b/include/configs/sunxi-common.h | 14 b/include/netdev.h |2 b/tools/mksunxiboot.c| 10 drivers/net/sunxi_wemac.c| 525 --- u-boot/arch/arm/cpu/armv7/sunxi/Makefile |4 u-boot/arch/arm/cpu/armv7/sunxi/board.c | 20 u-boot/arch/arm/cpu/armv7/sunxi/clock_sun4i.c|4 u-boot/arch/arm/cpu/armv7/sunxi/cpu_info.c | 15 u-boot/arch/arm/cpu/armv7/sunxi/dram.c | 102 +++ u-boot/arch/arm/include/asm/arch-sunxi/clock_sun4i.h | 11 u-boot/board/sunxi/Makefile |5 u-boot/board/sunxi/ahci.c| 84 +++ u-boot/board/sunxi/board.c | 32 + u-boot/board/sunxi/dram_a13_oli_micro.c | 32 + u-boot/board/sunxi/dram_cubieboard.c | 31 + u-boot/board/sunxi/dram_cubieboard2.c| 31 + u-boot/board/sunxi/dram_r7dongle.c | 31 + u-boot/boards.cfg| 25 u-boot/drivers/block/ahci.c | 16 u-boot/drivers/power/Makefile|2 u-boot/drivers/power/axp152.c| 112 u-boot/drivers/power/axp209.c| 180 ++ u-boot/include/ahci.h|4 u-boot/include/axp152.h | 11 u-boot/include/axp209.h | 15 u-boot/include/configs/sun4i.h | 24 u-boot/include/configs/sun5i.h | 24 u-boot/include/configs/sun7i.h |1 u-boot/include/configs/sunxi-common.h| 24 46 files changed, 1549 insertions(+), 692 deletions(-) Please confirm if it works for you, or have suggestions or an alternate patchset to apply. live well, vagrant signature.asc Description: Digital signature
Bug#750473: u-boot: Please enable Cubietruck support
On Thu, 2014-06-12 at 16:09 -0700, Vagrant Cascadian wrote: Definitely starting to think about splitting the u-boot package into one package per platform, at least for armhf... I'm not aware of your reasoning for that, but in my experience splitting packages up in very small pieces in not a great idea. Even if I don't doubt you're ability to deal with the maintenance overhead, you'll still hit the NEW queue alot Each additional platform adds .5+ MB to the package, and some of these platforms may have limited space, and little or no use for platforms other than their own... Slightly orthogonally, it would be useful (for FEL boot) to be able to install u-boot packages on foreign architectures (so e.g. I can install the cubietruck u-boot on the x86 box at the other end of the USB cable to my truck). I suppose multiarch is the answer rather than Arch:all packages? (Especially since each u-boot will want to be built natively, at least for now...) The size of the package will have nearly doubled since wheezy, and I wouldn't be surprised if people want to add more platforms. I think sunxi in particular will end up contributing *lots* of new platforms. But that's kind of outside the scope of this bug report... Yeah, sorry to add to the off-topicness. Ian. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
On Wed, Jun 04, 2014 at 07:54:08AM +0100, Ian Campbell wrote: On Tue, 2014-06-03 at 22:57 +0200, Andreas Henriksson wrote: On Tue, Jun 03, 2014 at 11:20:22AM -0700, Vagrant Cascadian wrote: On Tue, Jun 03, 2014 at 06:20:03PM +0200, Andreas Henriksson wrote: Add something like this to debian/targets: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin For extra bonus points also enable FEL mode, possibly like this: armhf Cubietruck_FEL u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin I didn't enable the _FEL variant, do you definitely need that as well? ... I haven't actually tried this yet but I was half thinking it might be nice/useful to be able to boot D-I from USB using FEL instead of messing around with TFTP or MMC cards etc. That would require the FEL versions of u-boot.bin and spl/sunxi-spl.bin (not sure abuot u-boot.img, I don't use it with FEL at least). ... FWIW FEL mode would want them split not joined. And I think you need u-boot.bin not u-boot.img for that case too (or at least that is what I've always used...). Ok, based on the above, I'm inclined to build with: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin armhf Cubietruck_FEL u-boot.bin spl/u-boot-spl.bin (spl/sunxi-spl.bin failed to build with the _FEL target) Hopefully those are the targets you need? If that sounds good, I'll include in the next upload. Definitely starting to think about splitting the u-boot package into one package per platform, at least for armhf... live well, vagrant signature.asc Description: Digital signature
Bug#750473: u-boot: Please enable Cubietruck support
Hello Vagrant Cascadian! Thanks for your work! Much appreciated. A few minor nitpicks/comments/feedback below/inline... On Thu, Jun 12, 2014 at 12:17:48PM -0700, Vagrant Cascadian wrote: [...] Ok, based on the above, I'm inclined to build with: I only tested u-boot-sunxi-with-spl.bin myself, but for consistency with Cubietruck_FEL and if I interpretted Ian Campbell correctly using u-boot.bin target would probably be preferred over u-boot.img in the below line: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin armhf Cubietruck_FEL u-boot.bin spl/u-boot-spl.bin (spl/sunxi-spl.bin failed to build with the _FEL target) Hopefully those are the targets you need? If that sounds good, I'll include in the next upload. Anyway, this is a very good start! Thanks! I think Ian Campbell also had a few good points about SATA support so if you haven't already considered it I would recommend looking at http://lists.denx.de/pipermail/u-boot/2014-June/181047.html (Hopefully these will be merged upstream any day now..) Definitely starting to think about splitting the u-boot package into one package per platform, at least for armhf... I'm not aware of your reasoning for that, but in my experience splitting packages up in very small pieces in not a great idea. Even if I don't doubt you're ability to deal with the maintenance overhead, you'll still hit the NEW queue alot Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
On Thu, Jun 12, 2014 at 09:39:33PM +0200, Andreas Henriksson wrote: On Thu, Jun 12, 2014 at 12:17:48PM -0700, Vagrant Cascadian wrote: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin armhf Cubietruck_FEL u-boot.bin spl/u-boot-spl.bin (spl/sunxi-spl.bin failed to build with the _FEL target) Hopefully those are the targets you need? If that sounds good, I'll include in the next upload. Not sure if the upstream patches are consistant with this, but: https://github.com/linux-sunxi/u-boot-sunxi/wiki Suggests that they've switched to u-boot-sunxi-with-spl.bin by default, and that u-boot.img replaced the default u-boot.bin. I'll leave out u-boot.img and spl/sunxi-spl.bin and someone can request them if there's a valid use-case. http://docs.cubieboard.org/tutorials/ct1/development/booting_from_usb_otg_port Suggests FEL mode needs u-boot.bin and u-boot-spl.bin, so that's what I'll include, so that's: armhf Cubietruck u-boot-sunxi-with-spl.bin armhf Cubietruck_FEL u-boot.bin spl/u-boot-spl.bin Seems to build fine that way, at least. I think Ian Campbell also had a few good points about SATA support so if you haven't already considered it I would recommend looking at http://lists.denx.de/pipermail/u-boot/2014-June/181047.html (Hopefully these will be merged upstream any day now..) Just waiting for upstream inclusion, unless they're simple and non-invasive. Definitely starting to think about splitting the u-boot package into one package per platform, at least for armhf... I'm not aware of your reasoning for that, but in my experience splitting packages up in very small pieces in not a great idea. Even if I don't doubt you're ability to deal with the maintenance overhead, you'll still hit the NEW queue alot Each additional platform adds .5+ MB to the package, and some of these platforms may have limited space, and little or no use for platforms other than their own... The size of the package will have nearly doubled since wheezy, and I wouldn't be surprised if people want to add more platforms. But that's kind of outside the scope of this bug report... live well, vagrant signature.asc Description: Digital signature
Bug#750473: u-boot: Please enable Cubietruck support
On Tue, 2014-06-03 at 22:57 +0200, Andreas Henriksson wrote: Hello Vagrant! Thanks for your extremely quick reply and backporting work! Ack! On Tue, Jun 03, 2014 at 11:20:22AM -0700, Vagrant Cascadian wrote: On Tue, Jun 03, 2014 at 06:20:03PM +0200, Andreas Henriksson wrote: Please consider shipping the package with Cubietruck support enabled. Cubietruck support has been merged upstream recently (post 2014.07-rc2). ... Packaging a git snapshot for experimental for now would be appreciated while waiting for the final v2014.07 release. Built a package with the patches from upstream applied on v2014.04 (was a fairly straightforward backport), and it built fine. Did you also pick up the designware fixes? (It sounds like it since Andreas reports dhcp working). FYI there is support for cubieboard 12 and a couple of other boards (one ofthe OLinuxIno ones I think) ongoing on the u-boot list. To be really useful we would also want the AHCI driver which is also currently being reviewed on the list. I suppose all of that ought to wait for upstream acceptance though. Add something like this to debian/targets: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin For extra bonus points also enable FEL mode, possibly like this: armhf Cubietruck_FEL u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin I didn't enable the _FEL variant, do you definitely need that as well? I'm not planning to use it myself right now, but others might very well need it and maybe I will too in the future. I see no reason not to enable it, but maybe you know better why not to. I haven't actually tried this yet but I was half thinking it might be nice/useful to be able to boot D-I from USB using FEL instead of messing around with TFTP or MMC cards etc. That would require the FEL versions of u-boot.bin and spl/sunxi-spl.bin (not sure abuot u-boot.img, I don't use it with FEL at least). That's part of my reason for #750104, to provide the necessary tools on the host side. (I haven't really considered how the u-boot would get onto the host system -- since the package is not arch:all. I suppose multiarch might help!) FEL booting is for example needed for USB booting. For more info on FEL see http://linux-sunxi.org/FEL and subpages like http://linux-sunxi.org/FEL/USBBoot (Note: building u-boot-sunxi-with-spl.bin should be enough really, other targets just included for following the general style and someone might find them useful...) If the only file needed is u-boot-sunxi-with-spl.bin, and it builds fine, then I'd be inclined to only ship that. Right now we only have MMC driver in the Debian kernel. The NAND driver is going through reviews on Linux kernel mailing list and will hopefully soon be merged (for 3.17 merge window maybe? If so, then we can probably get it into 3.16 kernel for Debian in ~ 2 months.) AFAIK normal booting is only available though MMC and NAND on the A20 SoC, so right now u-boot-sunxi-with-spl.bin is the most convenient to use. Fwiw, u-boot-sunxi-with-spl.bin is just a combination of the other two so they will still be built. Also maybe we might soon need them split out for NAND booting (but I haven't looked at what's needed for NAND booting yet). FWIW FEL mode would want them split not joined. And I think you need u-boot.bin not u-boot.img for that case too (or at least that is what I've always used...). I don't see a reason to not ship them (no compile time save and only minimal size difference on the binary package), but again you might know better. Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
Hello Ian! Thanks for your input! On Wed, Jun 04, 2014 at 07:54:08AM +0100, Ian Campbell wrote: [...] Did you also pick up the designware fixes? (It sounds like it since Andreas reports dhcp working). There where a bunch of designware patches in the package. FYI there is support for cubieboard 12 and a couple of other boards (one ofthe OLinuxIno ones I think) ongoing on the u-boot list. To be really useful we would also want the AHCI driver which is also currently being reviewed on the list. I suppose all of that ought to wait for upstream acceptance though. [...] Hopefully you'll be able to get these patches in before v2014.07 and I think we can live without AHCI until then (which AIUI means there will be a need for a /boot partition on MMC which isn't the end of the world). Of the unmerged patches on the mailinglist, this one looks like a fix for a pretty horrible problem though: http://lists.denx.de/pipermail/u-boot/2014-June/180936.html Maybe the description makes it sound worse then it really is? Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
On Wed, 2014-06-04 at 11:24 +0200, Andreas Henriksson wrote: Hello Ian! Thanks for your input! On Wed, Jun 04, 2014 at 07:54:08AM +0100, Ian Campbell wrote: [...] Did you also pick up the designware fixes? (It sounds like it since Andreas reports dhcp working). There where a bunch of designware patches in the package. FYI there is support for cubieboard 12 and a couple of other boards (one ofthe OLinuxIno ones I think) ongoing on the u-boot list. To be really useful we would also want the AHCI driver which is also currently being reviewed on the list. I suppose all of that ought to wait for upstream acceptance though. [...] Hopefully you'll be able to get these patches in before v2014.07 and I think we can live without AHCI until then (which AIUI means there will be a need for a /boot partition on MMC which isn't the end of the world). The problem is that in sid we currently have Linux v3.14 which speaks AHCI but not MMC and u-boot upstream currently speaks MMC but not AHCI. The Linux MMC patches are currently backported to our v3.15 kernel which is in experimental. But it has now transpired that v3.14 is likely to be the next upstream longterm stable [0], which has an impact on what will be in Jessie. If we end up going with v3.14 I'll have to investigate backporting the MMC patches further to 3.14, I'm not sure how plausible that will be. (Personally my use cases centre around SATA disks more than MMC, but I'm still happy to attempt the driver backport to 3.14 if it turns out to be needed) [0] https://lists.debian.org/debian-kernel/2014/06/msg00017.html Of the unmerged patches on the mailinglist, this one looks like a fix for a pretty horrible problem though: http://lists.denx.de/pipermail/u-boot/2014-June/180936.html Maybe the description makes it sound worse then it really is? It is pretty nasty, but AIUI it is benign unless you have other mainline patches which actually expose the issue, so unless we backport those (I don't know which they are, but I think we are unlikely to grab them) I think we are OK. Once that patch goes upstream we should definitely take it though. Ian. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
Source: u-boot Version: 2014.04-2 Severity: wishlist Dear Maintainer, Please consider shipping the package with Cubietruck support enabled. Cubietruck support has been merged upstream recently (post 2014.07-rc2). (Look for patches by by Ian Campbell if you'd rather considering backporting the commits instead of using a git snapshot). Work has already been done on the Debian kernel to make it usable on a headless Cubietruck and will be in the next experimental upload (which is imminent), see Bug#749484 for more information on this. Packaging a git snapshot for experimental for now would be appreciated while waiting for the final v2014.07 release. What is needed: = Merge a recent upstream git commit (ie. 3fe1a8). Drop or rebase patches which no longer applies (ie. cubox-i patches, other patches that are now included in new upstream). Add something like this to debian/targets: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin For extra bonus points also enable FEL mode, possibly like this: armhf Cubietruck_FEL u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin (Note: building u-boot-sunxi-with-spl.bin should be enough really, other targets just included for following the general style and someone might find them useful...) ... Profit! -- System Information: Debian Release: jessie/sid APT prefers testing-updates APT policy: (500, 'testing-updates'), (500, 'testing'), (1, 'experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.14-1-amd64 (SMP w/4 CPU cores) Locale: LANG=en_DK.UTF-8, LC_CTYPE=en_DK.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Bug#750473: u-boot: Please enable Cubietruck support
On Tue, Jun 03, 2014 at 06:20:03PM +0200, Andreas Henriksson wrote: Please consider shipping the package with Cubietruck support enabled. Cubietruck support has been merged upstream recently (post 2014.07-rc2). ... Packaging a git snapshot for experimental for now would be appreciated while waiting for the final v2014.07 release. Built a package with the patches from upstream applied on v2014.04 (was a fairly straightforward backport), and it built fine. I don't have a cubietruck, so if you could please test that it works: http://people.debian.org/~vagrant/u-boot/ Add something like this to debian/targets: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin For extra bonus points also enable FEL mode, possibly like this: armhf Cubietruck_FEL u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin I didn't enable the _FEL variant, do you definitely need that as well? (Note: building u-boot-sunxi-with-spl.bin should be enough really, other targets just included for following the general style and someone might find them useful...) If the only file needed is u-boot-sunxi-with-spl.bin, and it builds fine, then I'd be inclined to only ship that. live well, vagrant signature.asc Description: Digital signature
Bug#750473: u-boot: Please enable Cubietruck support
Hello Vagrant! Thanks for your extremely quick reply and backporting work! On Tue, Jun 03, 2014 at 11:20:22AM -0700, Vagrant Cascadian wrote: On Tue, Jun 03, 2014 at 06:20:03PM +0200, Andreas Henriksson wrote: Please consider shipping the package with Cubietruck support enabled. Cubietruck support has been merged upstream recently (post 2014.07-rc2). ... Packaging a git snapshot for experimental for now would be appreciated while waiting for the final v2014.07 release. Built a package with the patches from upstream applied on v2014.04 (was a fairly straightforward backport), and it built fine. I don't have a cubietruck, so if you could please test that it works: http://people.debian.org/~vagrant/u-boot/ I've done some very minimal testing. Installing the u-boot-sunxi-with-spl.bin from your package on to a microSD-card and booting it shows the u-boot prompt. After setting ethaddr I could retrieve an ip via dhcp and ping my gateway. Not tried actually booting a linux kernel yet. (3.15.0-rc8 is being uploaded to the archive as we speak... so I'm waiting for it to be accepted and built on armhf.) Add something like this to debian/targets: armhf Cubietruck u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin For extra bonus points also enable FEL mode, possibly like this: armhf Cubietruck_FEL u-boot.img spl/sunxi-spl.bin u-boot-sunxi-with-spl.bin I didn't enable the _FEL variant, do you definitely need that as well? I'm not planning to use it myself right now, but others might very well need it and maybe I will too in the future. I see no reason not to enable it, but maybe you know better why not to. FEL booting is for example needed for USB booting. For more info on FEL see http://linux-sunxi.org/FEL and subpages like http://linux-sunxi.org/FEL/USBBoot (Note: building u-boot-sunxi-with-spl.bin should be enough really, other targets just included for following the general style and someone might find them useful...) If the only file needed is u-boot-sunxi-with-spl.bin, and it builds fine, then I'd be inclined to only ship that. Right now we only have MMC driver in the Debian kernel. The NAND driver is going through reviews on Linux kernel mailing list and will hopefully soon be merged (for 3.17 merge window maybe? If so, then we can probably get it into 3.16 kernel for Debian in ~ 2 months.) AFAIK normal booting is only available though MMC and NAND on the A20 SoC, so right now u-boot-sunxi-with-spl.bin is the most convenient to use. Fwiw, u-boot-sunxi-with-spl.bin is just a combination of the other two so they will still be built. Also maybe we might soon need them split out for NAND booting (but I haven't looked at what's needed for NAND booting yet). I don't see a reason to not ship them (no compile time save and only minimal size difference on the binary package), but again you might know better. Regards, Andreas Henriksson -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org