Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 21 Apr 2025 18:56:13 +0100
From:      Jessica Clarke <jrtc27@freebsd.org>
To:        =?utf-8?Q?Bojan_Novkovi=C4=87?= <bnovkov@FreeBSD.org>
Cc:        "src-committers@freebsd.org" <src-committers@FreeBSD.org>, "dev-commits-src-all@freebsd.org" <dev-commits-src-all@FreeBSD.org>, "dev-commits-src-main@freebsd.org" <dev-commits-src-main@FreeBSD.org>
Subject:   Re: git: f1e18f331923 - main - riscv: Exclude OpenSBI memory regions when booting with EFI
Message-ID:  <F443E453-62F7-4599-A6B9-118FB8D63E69@freebsd.org>
In-Reply-To: <6F9CE6EB-8A7E-48D1-BDE5-AB7D930EC1DE@freebsd.org>
References:  <202504161421.53GELphv059586@gitrepo.freebsd.org> <6F9CE6EB-8A7E-48D1-BDE5-AB7D930EC1DE@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 16 Apr 2025, at 15:30, Jessica Clarke <jrtc27@FreeBSD.org> wrote:
>=20
> On 16 Apr 2025, at 15:21, Bojan Novkovi=C4=87 <bnovkov@FreeBSD.org> =
wrote:
>>=20
>> The branch main has been updated by bnovkov:
>>=20
>> URL: =
https://cgit.FreeBSD.org/src/commit/?id=3Df1e18f331923041980149fef46cdb273=
6e61debb
>>=20
>> commit f1e18f331923041980149fef46cdb2736e61debb
>> Author:     Bojan Novkovi=C4=87 <bnovkov@FreeBSD.org>
>> AuthorDate: 2025-04-15 16:28:05 +0000
>> Commit:     Bojan Novkovi=C4=87 <bnovkov@FreeBSD.org>
>> CommitDate: 2025-04-16 14:20:13 +0000
>>=20
>>   riscv: Exclude OpenSBI memory regions when booting with EFI
>>=20
>>   OpenSBI uses the first PMP entry to prevent buggy supervisor
>>   software from overwriting the firmware [1]. However, this
>>   region may not be properly marked as reserved in the EFI map, =
leading
>>   to an access violation exception whenever the kernel
>>   attempts to write to a page from that region.
>>=20
>>   Fix this by preemptively excluding first EFI memory map entry
>>   if it is marked as "BootServicesData".
>>=20
>>   [1] https://github.com/riscv-non-isa/riscv-sbi-doc/pull/37
>>=20
>>   Reported by: tuexen
>>   Tested by: tuexen
>>   Fixes: a2e2178402af
>>   Reviewed by:    imp, jrtc27
>=20
> No I didn=E2=80=99t, I left a comment saying I didn=E2=80=99t like the =
concept.
>=20
>>   Differential Revision:  https://reviews.freebsd.org/D49839
>> ---
>> sys/riscv/riscv/machdep.c | 32 ++++++++++++++++++++++++++++++--
>> 1 file changed, 30 insertions(+), 2 deletions(-)
>>=20
>> diff --git a/sys/riscv/riscv/machdep.c b/sys/riscv/riscv/machdep.c
>> index 516dbde5ffaa..f253bc9a853b 100644
>> --- a/sys/riscv/riscv/machdep.c
>> +++ b/sys/riscv/riscv/machdep.c
>> @@ -541,6 +541,22 @@ fdt_physmem_exclude_region_cb(const struct =
mem_region *mr, void *arg __unused)
>> }
>> #endif
>>=20
>> +static void
>> +efi_exclude_sbi_pmp_cb(struct efi_md *p, void *argp)
>> +{
>> + bool *first =3D (bool *)argp;
>> +
>> + if (!*first)
>> + return;
>> +
>> + *first =3D false;
>> + if (p->md_type =3D=3D EFI_MD_TYPE_BS_DATA) {
>> + physmem_exclude_region(p->md_phys,
>> +    min(p->md_pages * EFI_PAGE_SIZE, L2_SIZE),
>> +    EXFLAG_NOALLOC);
>=20
> Doesn=E2=80=99t this need EXFLAG_NODUMP like the FDT case?

Ping.

Jess




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?F443E453-62F7-4599-A6B9-118FB8D63E69>