On Mon, May 17, 2021 at 9:19 AM Bruce Ashfield <[email protected]> wrote: > > I've grabbed this change, but I wanted to give you a heads up that 5.10.37 has > created a lot of conflicts in the branch: > > both modified: arch/arm64/mm/init.c > both modified: drivers/clk/imx/clk-imx25.c > both modified: drivers/clk/imx/clk-imx27.c > both modified: drivers/clk/imx/clk-imx35.c > both modified: drivers/clk/imx/clk-imx5.c > both modified: drivers/clk/imx/clk-imx6q.c > both modified: drivers/clk/imx/clk-imx6sl.c > both modified: drivers/clk/imx/clk-imx6sll.c > both modified: drivers/clk/imx/clk-imx6sx.c > both modified: drivers/clk/imx/clk-imx7d.c > both modified: drivers/clk/imx/clk-imx7ulp.c > both modified: drivers/clk/imx/clk-imx8mm.c > both modified: drivers/clk/imx/clk-imx8mn.c > both modified: drivers/clk/imx/clk-imx8mp.c > both modified: drivers/clk/imx/clk-imx8mq.c > both modified: drivers/clk/imx/clk.c > both modified: drivers/clk/imx/clk.h > both modified: drivers/i2c/busses/i2c-imx.c > both modified: sound/soc/codecs/wm8960.c > > In particular it is the clock registration now takes a number of clocks, > versus > the boolean that was in the branch before. I've gone with the -stable version > of changes, but obviously, I don't have a board to test. > > I find it easier to take a small breakage window if the merge has gone bad, so > I'll push the changes, and you can revert and fix, or just send more patches > to > fix what is wrong on the branch.
I've changed strategies on this, because the changes are just different enough, that there was no way I'd get the merge correct (since I can't do a build-boot cycle). Instead, I've done: git merge -X ours v5.10/standard/base So the clock changes are intact from the sdk, but they will still be problems from the -stable merge. That should be an easier fixup for you. Bruce > > Bruce > > > > On Fri, May 14, 2021 at 3:39 AM Xiaolei Wang <[email protected]> > wrote: > > > > When porting the SDK commit 593bea4e36d8 ("MLK-16005-2 arm64: tlb: add > > the SW workaround for i.MX8QM TKT340553"), we neglected to apply the > > TKT340553 Errata workaround to the TLB non-range operations due > > to the variance between the SDK and our kernel. We have observed random > > page fault calltrace due to this. Fix it by applying the TKT340553 > > Errata workaround to the TLB non-range operations. > > > > Signed-off-by: Xiaolei Wang <[email protected]> > > --- > > arch/arm64/include/asm/tlbflush.h | 4 +++- > > 1 file changed, 3 insertions(+), 1 deletion(-) > > > > diff --git a/arch/arm64/include/asm/tlbflush.h > > b/arch/arm64/include/asm/tlbflush.h > > index 6218fa355149..a27a75fbe048 100644 > > --- a/arch/arm64/include/asm/tlbflush.h > > +++ b/arch/arm64/include/asm/tlbflush.h > > @@ -344,7 +344,9 @@ static inline void __flush_tlb_range(struct > > vm_area_struct *vma, > > if (!system_supports_tlb_range() || > > pages % 2 == 1) { > > addr = __TLBI_VADDR(start, asid); > > - if (last_level) { > > + if (TKT340553_SW_WORKAROUND) { > > + __tlbi(vmalle1is); > > + } else if (last_level) { > > __tlbi_level(vale1is, addr, tlb_level); > > __tlbi_user_level(vale1is, addr, tlb_level); > > } else { > > -- > > 2.25.1 > > > > > -- > - Thou shalt not follow the NULL pointer, for chaos and madness await > thee at its end > - "Use the force Harry" - Gandalf, Star Trek II -- - Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end - "Use the force Harry" - Gandalf, Star Trek II
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#9887): https://lists.yoctoproject.org/g/linux-yocto/message/9887 Mute This Topic: https://lists.yoctoproject.org/mt/82819195/21656 Group Owner: [email protected] Unsubscribe: https://lists.yoctoproject.org/g/linux-yocto/unsub [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
