Re: [PATCH v2] ACPI: sysfs: Change ACPI_MASKABLE_GPE_MAX to 0x100

[Rafael J. Wysocki]

On Tue, Nov 19, 2019 at 3:22 AM Yunfeng Ye <yeyunfeng@xxxxxxxxxx> wrote:
>
>
>
> On 2019/11/19 5:15, Rafael J. Wysocki wrote:
> > On Mon, Nov 18, 2019 at 8:44 PM Dexuan-Linux Cui <dexuan.linux@xxxxxxxxx> wrote:
> >>
> >> On Mon, Nov 18, 2019 at 1:04 AM Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote:
> >>>
> >>> On Sat, Nov 16, 2019 at 9:06 PM Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
> >>>>
> >>>> On Wed, Nov 13, 2019 at 11:17 PM Yunfeng Ye <yeyunfeng@xxxxxxxxxx> wrote:
> >>>>>
> >>>>> The commit 0f27cff8597d ("ACPI: sysfs: Make ACPI GPE mask kernel
> >>>>> parameter cover all GPEs") says:
> >>>>>   "Use a bitmap of size 0xFF instead of a u64 for the GPE mask so 256
> >>>>>    GPEs can be masked"
> >>>>>
> >>>>> But the masking of GPE 0xFF it not supported and the check condition
> >>>>> "gpe > ACPI_MASKABLE_GPE_MAX" is not valid because the type of gpe is
> >>>>> u8.
> >>>>>
> >>>>> So modify the macro ACPI_MASKABLE_GPE_MAX to 0x100, and drop the "gpe >
> >>>>> ACPI_MASKABLE_GPE_MAX" check. In addition, update the docs "Format" for
> >>>>> acpi_mask_gpe parameter.
> >>>>>
> >>>>> Fixes: 0f27cff8597d ("ACPI: sysfs: Make ACPI GPE mask kernel parameter cover all GPEs")
> >>>>> Signed-off-by: Yunfeng Ye <yeyunfeng@xxxxxxxxxx>
> >>>>> ---
> >>>>> v1 -> v2:
> >>>>>  - drop the "gpe > ACPI_MASKABLE_GPE_MAX" check
> >>>>>  - update the docs "Format" from <int> to <byte> for acpi_mask_gpe parameter
> >>>>>  - update the commit comment
> >>>>>
> >>>>>  Documentation/admin-guide/kernel-parameters.txt | 2 +-
> >>>>>  drivers/acpi/sysfs.c                            | 4 ++--
> >>>>>  2 files changed, 3 insertions(+), 3 deletions(-)
> >>>>
> >>>> Bisect flags commit eb09878e1301 "ACPI: sysfs: Change
> >>>> ACPI_MASKABLE_GPE_MAX to 0x100" in -next as the reason for a boot
> >>>> regression in my qemu-kvm test environment. It spews:
> >>>>
> >>>> [    1.456728] ACPI: Masking GPE 0x0.
> >>>> ...
> >>>> [  161.721420] ACPI: Masking GPE 0x0.
> >>>>
> >>>> ...and then hangs.
> >>>>
> >>>> A straight revert gets the configuration back on its feet.
> >>>>
> >>>> qemu-system-x86_64 --version
> >>>> QEMU emulator version 4.0.50 (v4.0.0-928-g49c6c6ac0cd8)
> >>>
> >>> OK, I'll drop it then, thanks!
> >>
> >> We're seeing the same issue wtih 5.4.0-rc7-next-20191118 on a Linux VM
> >> running on Hyper-V :
> >>
> >> [    0.133029] ACPI: 1 ACPI AML tables successfully acquired and loaded
> >> [    0.144023] ACPI: Interpreter enabled
> >> [    0.145023] ACPI: (supports S0 S5)
> >> [    0.146023] ACPI: Using IOAPIC for interrupt routing
> >> [    0.147024] PCI: Using host bridge windows from ACPI; if necessary,
> >> use "pci=nocrs" and report a bug
> >> [    0.148031] ACPI: Masking GPE 0x0.
> >> ...
> >> [  774.839023] ACPI: Masking GPE 0x0.
> >> [  774.840023] ACPI: Masking GPE 0x0.
> >>
> >> I guess the patch is only tested on a physical machine and not on a VM...
> >
> > It looks like the patch hasn't been tested at all.
> >
> > Please try to change the data type of gpe in
> > acpi_gpe_apply_masked_gpes() to u16 and see if that helps.
> >
> Sorry for this problem, I have no good test after modification according to
> inspection opinions.

I see.

Well, reviewers make mistakes too ...

> The first version patch is ok, the type of gpe is changed to u32. and I have
> test it before.

OK

So I have added the u8 -> u16 change for gpe in
acpi_gpe_apply_masked_gpes() and applied it again, as that should work
AFAICS.