[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <1767879.8hzESeGDPO@diego>
Date: Wed, 08 Jun 2022 14:23:32 +0200
From: Heiko Stübner <heiko@...ech.de>
To: Palmer Dabbelt <palmer@...osinc.com>, Ron Economos <re@...z.net>
Cc: linux-riscv@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [GIT PULL] RISC-V Patches for the 5.19 Merge Window, Part 1
Am Mittwoch, 8. Juni 2022, 02:12:47 CEST schrieb Ron Economos:
> On 6/7/22 4:36 PM, Heiko Stübner wrote:
> > Am Dienstag, 7. Juni 2022, 22:46:52 CEST schrieb Ron Economos:
> >> On 5/31/22 10:13 AM, Palmer Dabbelt wrote:
> >>> The following changes since commit 9282d0996936c5fbf877c0d096a3feb456c878ad:
> >>>
> >>> csky: Move to generic ticket-spinlock (2022-05-11 11:50:15 -0700)
> >>>
> >>> are available in the Git repository at:
> >>>
> >>> git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux.git tags/riscv-for-linus-5.19-mw0
> >>>
> >>> for you to fetch changes up to 7699f7aacf3ebfee51c670b6f796b2797f0f7487:
> >>>
> >>> RISC-V: Prepare dropping week attribute from arch_kexec_apply_relocations[_add] (2022-05-30 16:04:37 -0700)
> >>>
> >>> ----------------------------------------------------------------
> >>> RISC-V Patches for the 5.19 Merge Window, Part 1
> >>>
> >>> * Support for the Svpbmt extension, which allows memory attributes to be
> >>> encoded in pages.
> >>>
> >>>
> >>> Heiko Stuebner (12):
> >>> riscv: integrate alternatives better into the main architecture
> >>> riscv: allow different stages with alternatives
> >>> riscv: implement module alternatives
> >>> riscv: implement ALTERNATIVE_2 macro
> >>> riscv: extend concatenated alternatives-lines to the same length
> >>> riscv: prevent compressed instructions in alternatives
> >>> riscv: move boot alternatives to after fill_hwcap
> >>> riscv: Fix accessing pfn bits in PTEs for non-32bit variants
> >>> riscv: add RISC-V Svpbmt extension support
> >>> riscv: remove FIXMAP_PAGE_IO and fall back to its default value
> >>> riscv: don't use global static vars to store alternative data
> >>> riscv: add memory-type errata for T-Head
> >>>
> >> An issue was found on the HiFive Unmatched with the 5.19-rc1 kernel. The
> >> following warning occurs during boot:
> >>
> >> riscv64 kernel:
> >> ----------------------------------------------------------------
> >> riscv64 kernel: WARNING: Missing the following errata may cause
> >> potential issues
> >> riscv64 kernel: SiFive Errata[0]:cip-453
> >> riscv64 kernel: SiFive Errata[1]:cip-1200
> >> riscv64 kernel: Please enable the corresponding Kconfig to apply them
> >> riscv64 kernel:
> >> ----------------------------------------------------------------
> >>
> >> I've manually bisected the problem to this commit:
> >>
> >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=ff689fd21cb13098305bae3f8d0c0065df2e2fc1
> > hmm, on first glance I can't really see how that specific commit would
> > affect Sifive erratas. Patches directly before this one did change
> > alternative / errata things though.
> >
> > (1) From looking at your output, it seems at least CONFIG_ERRATA_SIFIVE
> > is enabled as the general sifive errata code is running, but are the
> > individual erratas also still enabled in your config?
> > They're default=y forever and weren't touched by the svpbmt series,
> > but it'd be nice to know how the config looks, to see if the Kconfig
> > settings are strange somewhere
> >
> >
> > (2) Going from the list on
> >
> > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/log/?h=ff689fd21cb13098305bae3f8d0c0065df2e2fc1
> >
> > could you maybe check if it's really working with the parent of the
> > commit you found. (i.e. patches before the svpbmt addition).
> >
> > Also just to be sure, it still works correctly with 5.18, right?
> >
> >
> > Thanks
> > Heiko
> >
> >
> Yes, that commit was surprising. I also thought it would have been one
> of the previous "alternatives" commits. But I tested them all, and the
> warning only appears with the "add RISC-V Svpbmt extension support"
> commit. In other words, "git checkout 100631b" works fine, but "git
> checkout ff689fd" does not.
after talking with Ron a bit more, we came across the finding that the
warnings actually happen later in the boot process, pointing to modules.
Which lead me to [0] for explanation of the issue and a possible fix.
[0] https://lore.kernel.org/r/20220608120849.1695191-1-heiko@sntech.de/
Powered by blists - more mailing lists