Hi Greg,
For your consideration following upstream patches taken from lede
source tree[1] targeted for 4.9.y.
09f3510 (ARM: BCM5301X: Add back handler ignoring external)
5d1f2d2 (ARM: dts: BCM5301X: Set 5 GHz wireless frequency limits)
0c2bf9f (ARM: dts: BCM5301X: Correct GIC_PPI interrupt flags)
6e347b5 (PCI: iproc: Save host bridge window resource in struct)
6c356ed (MIPS: Lantiq: Fix cascaded IRQ setup)
e247454 (i2c: bcm2835: Fix hang for writing messages larger than 16 bytes)
d4030d7 (i2c: bcm2835: Protect against unexpected TXW/RXR interrupts)
23c9540 (i2c: bcm2835: Use dev_dbg logging on transfer errors)
8d2cc5cc (i2c: bcm2835: Can't support I2C_M_IGNORE_NAK)
155e8b3 (clk: bcm: Support rate change propagation on bcm2835 clocks)
d86d46a (clk: bcm: Allow rate change propagation to PLLH_AUX on VEC clock)
2aab7a2 (clk: bcm: Fix 'maybe-uninitialized' warning in
bcm2835_clock_choose_div_and_prate())
5548609 (clk: bcm2835: Don't rate change PLLs on behalf of DSI PLL dividers.)
8a39e9f (clk: bcm2835: Register the DSI0/DSI1 pixel clocks.)
3f91958 (clk: bcm2835: Add leaf clock measurement support, disabled by default)
2201ac6 (dmaengine: bcm2835: Fix cyclic DMA period splitting)
cd4b1e3 (usb: dwc2: Remove unnecessary kfree)
bd5d213 (mtd: bcm47xxpart: fix parsing first block after aligned TRX)
40be0dd (net: add devm version of alloc_etherdev_mqs function)
34a5102 (net: bgmac: allocate struct bgmac just once & don't copy it)
aa8863e (net: bgmac: drop struct bcma_mdio we don't need anymore)
3ec7544 (of: Add check to of_scan_flat_dt() before accessing
initial_boot_params)
7272416 (rt2500usb: don't mark register accesses as inline)
f4737a6 (brcmfmac: check brcmf_bus_get_memdump result for error)
36401cb (brcmfmac: be more verbose when PSM's watchdog fires)
9587a01 (brcmfmac: merge two brcmf_err macros into one)
087fa71 (brcmfmac: switch to C function (__brcmf_err) for printing errors)
d063055 (brcmfmac: merge two remaining brcmf_err macros)
93c7018 (rt2x00usb: do not anchor rx and tx urb's)
0488a61 (rt2x00usb: fix anchor initialization)
91b6328 (brcmfmac: Use net_device_stats from struct net_device)
a083c8f (rt2x00: Fix incorrect usage of CONFIG_RT2X00_LIB_USB)
6232c17 (rt2x00: avoid introducing a USB dependency in the rt2x00lib module)
Cherry-picked and build tested on Linux 4.9.20 for
ARCH=arm/arm64/i386/x86_64 + allmodconfig.
Regards,
Amit Pundir
[1] https://github.com/lede-project/source. These patches are mostly
cherry-picked from target/linux/*/patches-4.9 directories. Though some
are also put in package/kernel/*/patches/ directory.
Hi Greg, all,
Just a status: so I went through about 200+ patches since the last
patch list I sent; out of which 16 have been sent out to stable just
now.
8 listed below might be interesting but either don't apply easily or
need more understanding - if anyone finds them worthy for more time, I
can spend some, otherwise will drop them from my list
hv_netvsc: Add query for initial physical link speed
hv_netvsc: Add handler for physical link speed change
hv_netvsc: Implement batching of receive completions
netvsc: fix checksum on UDP IPV6
netvsc: Remove mistaken udp.h inclusion.
mm, oom: protect !costly allocations some more
mm: consider compaction feedback also for costly allocation
kvm: x86: correctly reset dest_map->vector when restoring LAPIC state
Best,
Sumit.
--
Thanks and regards,
Sumit Semwal
Linaro Mobile Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
Hi Greg,
I'm in the middle of cherry-picking and testing a handful of lede
https://lede-project.org stable patches as likely 4.9.y candidates. I
have few questions before I send them your way following Sasha's email
format.
1. If I use standard "git cherry-pick -x $commitID" to append original
git commit IDs as "(cherry picked from commit $commitID)" line in the
original commit message, will that work for you?
2. While verifying that these lede patches are indeed upstreamed
alrready, I ran into few more similar upstream fixes which though are
not part of lede distro but reasonable enough to go into stable-4.9.
Should I send them as well? May be as part of different set and not as
lede's patches?
Regards,
Amit Pundir
Hi Greg,
Apologies on the delay since last email; here are some patches from
ubuntu xenial 4.4 for your consideration towards stable-4.4.y.
cc: stable
ab2a4bf83902c170d29ba130a8abb5f9d90559e1 USB: don't free
bandwidth_mutex too early
c538b9436751a0be2e1246b48353bc23156bdbcc pinctrl: intel: Only restore
pins that are used by the driver
Seem useful; similar patches already in stable
3a020a723c65eb8ffa7c237faca26521a024e582 crypto: ghash-clmulni - Fix
load failure
1a07834024dfca5c4bed5de8f8714306e0a11836 crypto: cryptd - Assign
statesize properly
ddef482420b1ba8ec45e6123a7e8d3f67b21e5e3 crypto: mcryptd - Fix load failure
Seem useful:
83430833b4d4a9c9b23964babbeb1f36450f8136 cxlflash: Increase
cmd_per_lun for better throughput
e34fbbac669de0b7fb7803929d0477f35f6e2833 ACPI / video: skip evaluating
_DOD when it does not exist
bcb48cca23ec9852739e4a464307fa29515bbe48 pinctrl: cherryview: Do not
mask all interrupts in probe
77c0c9735bc0ba5898e637a3a20d6bcb50e3f67d Drivers: hv: balloon: don't
crash when memory is added in non-sorted order
a9f61ca793becabdefab03b77568d6c6f8c1bc79 Drivers: hv: avoid vfree() on crash
707e59ba494372a90d245f18b0c78982caa88e48 xen/qspinlock: Don't kick CPU
if IRQ is not initialized
708e75a3ee750dce1072134e630d66c4e6eaf63c KVM: PPC: Book3S PR: Fix
illegal opcode emulation
dba599091c191d209b1499511a524ad9657c0e5a s390/pci: fix use after free
in dma_init
A few more 'maybe's' that aren't marked cc: stable, but seemed useful;
I had lesser confidence in these, so listing them separate from the
above 'seem useful' ones
e9c5e7402dad6f4f04c2430db6f283512bcd4392 drm/amdgpu: add missing irq.h include
=> seems useful for building amdgpu on ARM
727f28b8ca24a581c7bd868326b8cea1058c720a tpm_tis: Use devm_free_irq not free_irq
=> simple fix
757647e10e55c01fb7a9c4356529442e316a7c72 hv_netvsc: use skb_get_hash()
instead of a homegrown implementation
=> fixes performance regression
4aec6ec0da9c72c0fa1a5b0d1133707481347bb3 cxl: Increase timeout for
detection of AFU mmio hang
=> helps detect mmio hang
725fc629ff2545b061407305ae51016c9f928fce kernek/fork.c: allocate idle
task for a CPU always on its local node
=> performance improvement?
cd95aad5579371ac332507fc946008217fc37e6c Drivers: hv: vmbus: handle
various crash scenarios
=> avoid kernel crashes
Hope these make some sense; with these, I've scanned through about
2300 patches till Jan'17 (Ubuntu-4.4.0-62.83 tag). I will continue to
do this for the changes since then.
Best regards,
Sumit
--
Thanks and regards,
Sumit Semwal
Linaro Mobile Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
Hi all,
It would be great if someone could go through the 4.9-lts patches that
Sasha posted to the stable@vger list last week and I included in the 4.9
release today, to see if anyone of them are also applicable to 4.4?
Also, the way he sent those patches out is a wonderful format to copy,
if you all want to do so (hint hint...)
thanks,
greg k-h
Hi Greg,
I found following (maybe)useful upstream cherry picks on
raspberrypi/rpi-4.9.y tree[1].
26fc78f6fef3 ("drm/vc4: Fix race between page flip completion event
and clean-up")
ab8df60e3a3b ("drm/vc4: Fix ->clock_select setting for the VEC encoder")
f2a46926aba1 ("clk: bcm2835: Fix ->fixed_divider of pllh_aux")
Cherry-picked and build tested on Linux 4.9.16 for
ARCH=arm/arm64/x86_64 + allmodconfig.
So far I'm relying on assumption that RPi maintainers use "git
cherry-pick -x" to cherry pick upstream commit on their stable trees.
I'll take a look again if I have missed out on any other patches.
Regards,
Amit Pundir
[1] https://github.com/raspberrypi/linux.git
Hi Greg,
I had a brief look at Fedora and a discussion with Fedora, I realised
she's already doing a decent job at sending you patches that might've
been missed from stable.
So started looking at Ubuntu 4.4 (xenial), and could find some patches
that seem good candidates for stable.
I'm just listing them down with upstream commit ID, description and a
remark, but if you prefer another way (txt file, xls, or anything
else), please let me know, and we can start following that :)
(This is just something to begin with, not done the whole list of
candidate patches. I will continue to monitor and send).
Patches cc'ed to stable:
91a25e463130c8e19bdb42f2d827836c7937992e drm/dp/mst: deallocate
payload on port destruction
8a580f70f6936ec095da217018cdeeb5835c0207 x86/irq: Do not use
apic_chip_data.old_domain as temporary buffer
3716fd27a604d61a91cda47083504971486b80f1 x86/irq: Check vector allocation early
8244062ef1e54502ef55f54cced659913f244c3e modules: fix longstanding
/proc/kallsyms vs module insertion race.
Patches not CC to stable, but seem a good fix
2b2f5ff00f63847d95adad6289bd8b05f5983dd5 rtc: interface: ignore
expired timers when enqueuing new timers
3084558658c2f9a48d7c460d57aeb30964c06b7e megaraid_sas: Don't issue
kill adapter for MFI controllers in case of PD list DCMD failure
=> fixes 3084558658c2f9a48d7c460d57aeb30964c06b7e
87c279e613f848c691111b29d49de8df3f4f56da blk-mq: really fix plug list
flushing for nomerge queues
=>fixes 0809e3ac6231
Also wanted to ask you if there's a better way in your opinion than
using "git cherry -v" to :
1. find patches IN distro, and IN mainline,
2. get patches IN distro, and NOT IN stable
and then scan through them to scout? :)
(If you're able to share any of your scripts for hunting distros, it'd
be pretty good I suppose!)
--
Thanks and regards,
Sumit Semwal
Linaro Mobile Group - Kernel Team Lead
Linaro.org │ Open source software for ARM SoCs
Hi All,
As a sort of 'prototype`' email for reporting back on one of the
potential sources for 4.9 kernel fixes, I went trolling into the
gentoo repo. (gentoo-sources to be exact, I didn't look at hardened
tho I will)
For 4.9 it's a pretty short list of potential fixes. Being more
verbose about this just because well first post as well get started,
here's what I've whittled the list down to and recommendations for the
two patches.
Patch: 1510_fs-enable-link-security-restrictions-by-default.patch
From: http://sources.debian.net/src/linux/3.16.7-ckt4-3/debian/patches/debian/fs-…
Desc: Enable link security restrictions by default.
Recommendation : Interesting but I have hard accepting this as a 'fix'
as compared to a feature change.
Patch: 2900_dev-root-proc-mount-fix.patch
From: https://bugs.gentoo.org/show_bug.cgi?id=438380
Desc: Ensure that /dev/root doesn't appear in /proc/mounts when
bootint without an initramfs.
Discussion: https://patchwork.kernel.org/patch/2076031/ (unresolved)
Recommendation: no action
--
Regards,
Tom
Director, Linaro Mobile Group
Linaro.org │ Open source software for ARM SoCs
irc: tgall_foo | skype : tom_gall
"Where's the kaboom!? There was supposed to be an earth-shattering
kaboom!" Marvin Martian
