#asahi · 2021-10-05 08:19
arndalso, regarding the timing: try to get an initial pull request out to soc@kernel.org early (i.e. before rc4 or rc5) if you have a lot of contents, you can always follow up with fixes or later contents when you have more stuff to add, I just don't want to get a large pull request shortly before the merge window as I'm trying to wrap stuff up for the pull requests I send to torvalds#asahi · 2021-10-05 08:16
arndbtw, if you update the email addresses in that file (new reviewer, change of employer etc), please send that directly as a bugfix for the current release so it has the correct information at the time of release, don't wait for the merge window with those#asahi · 2021-10-05 08:15
arndyes#asahi · 2021-10-05 08:14
arndwhen I get a pull request at soc@kernel.org, I usually take a quick look at all the contents to make sure nobody screwed up, but I expect the contents to have been reviewed at that point#asahi · 2021-10-05 08:13
arndright, so if you add an apple-m2.dtsi file that is not just adding support for another driver, that would probably need separate review, and you should loop in reviewers for that directly.#asahi · 2021-10-05 08:10
arndmarcan: for dts changes, I expect the review to happen during the review for the corresponding subsystem driver, and/or on the devicetree mailing list, cc linux-arm-kernel and anyone you want to participate in the review#asahi · 2021-10-05 08:09
arndif you only have a couple of changes for a given merge window, you can also just send individual patches instead of pull requests, the main purpose of the PR is to have a nicer git history with a merge commit that gets your tag description with an overview of the contents#asahi · 2021-10-05 08:08
arndmost of the changes in that tree would be for arch/arm64/boot/dts, but occasionally there could be something else, like Kconfig, defconfig, or MAINTAINERS updates#asahi · 2021-10-05 08:07
arndmarcan: the way this should normally work is that you have a tree to collect all M1 specific changes that don't go through a driver subsystem tree. This tree should be part of linux-next, and then you send a pull request to soc@kernel.org for me or ojn to pick it up into the soc tree#asahi · 2021-10-05 08:04
arndmaz: ideally that should go through marcan's tree, but if he has no other changes, I can take it from you with his ack#asahi · 2021-10-05 07:28
arndthe 'tip' tree works in a similar way, but they have a lot more branches#asahi · 2021-10-05 07:28
arndfor the soc tree, we have around five branches that are updated with individual pull requests and then merged into the for-next branch, but then the separate branches are what we send in the merge window#asahi · 2021-10-05 07:27
arndboth are merged into linux-next, but the fixes is for the current release, and may occasionally get merged into the for-next branch, which is sent in the following merge window#asahi · 2021-10-05 07:26
arndI think most maintainers have a 'fixes' branch and a 'for-next' branch, whatever they are named#asahi · 2021-08-14 17:50
arndmarcan: my original version skipped that part, I think sven added the obvious properties#asahi · 2021-08-09 17:49
arndthe main advantage of the mailbox framework is that there is an existing DT binding to describe how the various endpoints related to the shared mailbox driver#asahi · 2021-08-09 17:48
arndthe mailbox framework is a little strange in the way it works, but for this particular use case it actually sounds like a good fit to me, just exposing a fixed-length message fifo#asahi · 2021-08-09 16:04
arndsven: what is the mailbox needed for in case of nvme?#asahi · 2021-08-09 15:54
arndI guess you can start with a copy of the code you already have, and then remove everything that isn't currently getting used.#asahi · 2021-08-09 15:54
arndsven: well, duplicating that nvme driver at least makes it easier to get the initial version merged#asahi · 2021-08-09 14:44
arndor the hardware designers all read the same book on the topic#asahi · 2021-08-09 14:43
arndI just looked at all the other spi drivers to see if there was a closer match: it looks like half the drivers in the kernel have roughly the same design, with config/status/transmit/receive registers in that order, but they all have them at slightly different offsets, and with slightly different bit definitions, so the similarity to the s3c driver may just be coincidence#asahi · 2021-08-09 14:21
arndIf someone wants to push corellium's driver, that might actually work in this case, it seems well written, and fairly simple, much smaller than the two samsung drivers that implement DMA mode#asahi · 2021-08-09 14:20
arndmarcan: for the spi driver, I see some resemblance between its registers and both the s3c24xx and s3c64xx drivers, but it does seem different enough to need a third driver.#asahi · 2021-08-09 12:24
arndmarcan: it's not important to me to have it in there now, I'm mainly interested in not getting a large pull request just before the merge window#asahi · 2021-08-09 12:21
arndright, even if we had basic functionality in the kernel, there is not much point if distros don't come with an working installation method for the machine#asahi · 2021-08-09 12:16
arndhttp://phb-crystal-ball.org/ tells me that 5.15 is the next LTS kernel that next year's distros will ship#asahi · 2021-08-09 12:15
arndOk, good. Having USB working would already turn it from barely working into something one can actually play around with on an upstream kernel. I was originally hoping we could get PCI and NVMe in 5.15, but since NVMe certainly isn't getting in this time, I guess it will be 5.16 at the earliest for a mostly functioning system#asahi · 2021-08-09 11:23
arndmarcan: do you have any DT updates for 5.15 that you can send? This week would be great so I don't have to queue it into a 'late' branch. I haven't kept track of which drivers got merged since the original version that would need DT nodes#asahi · 2021-08-09 09:41
arndsven: joro just merged the dart patches for 5.15 \o/#asahi · 2021-08-06 15:47
arndnice!#asahi · 2021-08-06 15:45
arndanyway, if the gpu can see the cache, there is no need for write-through or uncached write-combining mapping of regular memory#asahi · 2021-08-06 15:44
arndoh, I wonder why that's different from arm32#asahi · 2021-08-06 15:43
arndthere is no need to ioremap() here, since it's just normal memory, not mmio#asahi · 2021-08-06 15:43
arndright#asahi · 2021-08-06 15:42
arndmarcan: You are probably thinking of ioremap_wc() vs ioremap(), which is the right idea for frame buffers to speed up consecutive stores#asahi · 2021-08-06 15:39
arndotherwise as marcan says, don't use that#asahi · 2021-08-06 15:38
arndIIRC dma_alloc_noncoherent() is a hack that is used for some of the old pa-risc/mips/ia64 numa systems, it has nothing to do with x86#asahi · 2021-08-06 08:21
arndpipcet[m]: I meant the Linux guest would either get the PCI XHCI (type-A on mac mini) ports, or the DCW3 (type-C) ports, or do you mean some bits are even shared between the two sets?#asahi · 2021-08-06 08:18
arndThere are a number of implementations along that line, in Linux we have 'jailhouse', and Intel is working on 'ACRN'#asahi · 2021-08-06 08:14
arndI don't see a real practical application for that kind of setup though, so I don't see who would spend the effort to implement that#asahi · 2021-08-06 08:13
arndOne thing that may be possible is to have a purely partitioning hypervisor that passes through (almost) everything to macos, while giving Linux access to half the CPUs, memory and one USB controller, and then put all the peripherals for the Linux guest on a USB hub.#asahi · 2021-07-29 15:16
arndIIRC most others that build their own chips (spreadtrum->unisoc, samsung, via->intel, infinion->intel->apple, hisilicon, mediatek) license the DSP plus 4g/5g software stack from CEVA#asahi · 2021-07-29 15:09
arndI thought quectel just uses qualcomm chips#asahi · 2021-07-29 15:06
arndAriadne: I don't think Intel has any modem development left, maybe they just retained the rights to keep selling the old SKUs to existing customers?#asahi · 2021-07-26 14:01
arndmy patches do this with a set of fairly minimal changes to the pci-nvme backend, to simplify rebasing across kernel versions, but it's possible that the maintainers want a more thorough split\#asahi · 2021-07-26 14:00
arndthe DT binding is the easy part in this case, splitting up the driver in a sensible way is slightly tricky#asahi · 2021-06-16 10:15
arndj_ey: that looks like a scary bug on the i.MX, I hope it's an SoC specific issue rather than a fundamental issue in the way we think of the barriers#asahi · 2021-06-10 14:50
arnddrivers/iommu/iommu-sysfs.c tells me there should be something in /sys/class/iommu/, but all the x86 machines I'm looking at have nothing in there, so maybe they just have it disabled#asahi · 2021-06-10 14:46
arndGlanzmann: not sure. Usually you can find some output in dmesg, but there may be a generic interface for querying at runtime that I'm not aware of