From: Zong Li <zong.li@sifive.com>
To: Alex Ghiti <alex@ghiti.fr>
Cc: Palmer Dabbelt <palmer@dabbelt.com>,
Paul Walmsley <paul.walmsley@sifive.com>,
linux-riscv <linux-riscv@lists.infradead.org>,
"linux-kernel@vger.kernel.org List"
<linux-kernel@vger.kernel.org>
Subject: Re: [PATCH RFC 6/8] riscv/kaslr: clear the original kernel image
Date: Tue, 7 Apr 2020 19:18:12 +0800 [thread overview]
Message-ID: <CANXhq0p7A2HeyFDRQaw5brzembsFM0-v3kPwQKvFZuyeCm6tsg@mail.gmail.com> (raw)
In-Reply-To: <772ee8e0-f5ff-cf40-4e84-3f703953cd08@ghiti.fr>
On Tue, Apr 7, 2020 at 1:11 PM Alex Ghiti <alex@ghiti.fr> wrote:
>
> On 3/24/20 3:30 AM, Zong Li wrote:
> > After completing final page table, we can clear original kernel image
> > and remove executable permission.
> >
> > Signed-off-by: Zong Li <zong.li@sifive.com>
> > ---
> > arch/riscv/include/asm/kaslr.h | 12 ++++++++++++
> > arch/riscv/kernel/kaslr.c | 12 ++++++++++++
> > arch/riscv/mm/init.c | 6 ++++++
> > 3 files changed, 30 insertions(+)
> > create mode 100644 arch/riscv/include/asm/kaslr.h
> >
> > diff --git a/arch/riscv/include/asm/kaslr.h b/arch/riscv/include/asm/kaslr.h
> > new file mode 100644
> > index 000000000000..b165fe71dd4a
> > --- /dev/null
> > +++ b/arch/riscv/include/asm/kaslr.h
> > @@ -0,0 +1,12 @@
> > +/* SPDX-License-Identifier: GPL-2.0-only */
> > +/*
> > + * Copyright (C) 2020 SiFive
> > + * Copyright (C) 2020 Zong Li <zong.li@sifive.com>
> > + */
> > +
> > +#ifndef _ASM_RISCV_KASLR_H
> > +#define _ASM_RISCV_KASLR_H
> > +
> > +void __init kaslr_late_init(void);
> > +
> > +#endif /* _ASM_RISCV_KASLR_H */
> > diff --git a/arch/riscv/kernel/kaslr.c b/arch/riscv/kernel/kaslr.c
> > index 59001d6fdfc3..0bd30831c455 100644
> > --- a/arch/riscv/kernel/kaslr.c
> > +++ b/arch/riscv/kernel/kaslr.c
> > @@ -356,6 +356,18 @@ static __init uintptr_t get_random_offset(u64 seed, uintptr_t kernel_size)
> > return get_legal_offset(random, kernel_size_align);
> > }
> >
> > +void __init kaslr_late_init(void)
> > +{
> > + uintptr_t kernel_size;
> > +
> > + /* Clear original kernel image. */
> > + if (kaslr_offset) {
> > + kernel_size = (uintptr_t) _end - (uintptr_t) _start;
>
> kernel_size = (uintptr_t) _end - (uintptr_t) _start + 1;
OK, change it in the next version. Thanks.
>
> > + memset((void *)PAGE_OFFSET, 0, kernel_size);
>
> I have been thinking again about our discussion regarding PAGE_OFFSET:
> PAGE_OFFSET actually points to the address where the kernel was loaded,
> not the beginning of memory, that's a bit weird.
>
> Just saying that here, because it took me a few seconds to remember that
> and understand what you were doing here.
In non-kaslr case, we load the kernel to PAGE_OFFSET which points to,
so we clear the old kernel image through PAGE_OFFSET here. Certainly,
we could use a symbol to record the start address of the old kernel
image instead of PAGE_OFFSET here. I don't see other architectures
changing PAGE_OFFSET after copying the kernel to the new location in
kaslr. If you think the PAGE_OFFSET needs to be changed, we need to
give another way to make the page table could create the mappings for
the whole memory and memblock/buddy system could see the whole memory
after the kernel moves.
>
> > + set_memory_nx(PAGE_OFFSET, kaslr_offset >> PAGE_SHIFT);
>
> Again, I certainly missed something but when do you use old kernel
> mappings ?
We use old kernel mappings when KASLR calculates the random offset, at
that moment, kernel is running on old kernel location.
>
> > + }
> > +}
> > +
> > uintptr_t __init kaslr_early_init(void)
> > {
> > u64 seed;
> > diff --git a/arch/riscv/mm/init.c b/arch/riscv/mm/init.c
> > index 34c6ecf2c599..08e2ce170533 100644
> > --- a/arch/riscv/mm/init.c
> > +++ b/arch/riscv/mm/init.c
> > @@ -15,6 +15,7 @@
> > #include <linux/set_memory.h>
> > #ifdef CONFIG_RELOCATABLE
> > #include <linux/elf.h>
> > +#include <asm/kaslr.h>
> > #endif
> >
> > #include <asm/fixmap.h>
> > @@ -649,6 +650,11 @@ static void __init setup_vm_final(void)
> > /* Move to swapper page table */
> > csr_write(CSR_SATP, PFN_DOWN(__pa_symbol(swapper_pg_dir)) | SATP_MODE);
> > local_flush_tlb_all();
> > +
> > +#ifdef CONFIG_RANDOMIZE_BASE
> > + /* Clear orignial kernel image and set the right permission. */
> > + kaslr_late_init();
> > +#endif
> > }
> >
> > void free_initmem(void)
> >
>
> Alex
next prev parent reply other threads:[~2020-04-07 11:18 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-03-24 7:30 [PATCH RFC 0/8] Support KASLR for RISC-V Zong Li
2020-03-24 7:30 ` [PATCH RFC 1/8] riscv/kaslr: add interface to get kaslr offset Zong Li
2020-04-07 5:08 ` Alex Ghiti
2020-03-24 7:30 ` [PATCH RFC 2/8] riscv/kaslr: introduce functions to clear page table Zong Li
2020-04-07 5:09 ` Alex Ghiti
2020-04-07 9:17 ` Zong Li
2020-03-24 7:30 ` [PATCH RFC 3/8] riscv/kaslr: support KASLR infrastructure Zong Li
2020-04-07 5:10 ` Alex Ghiti
2020-04-07 10:34 ` Zong Li
2020-04-09 5:53 ` Alex Ghiti
2020-04-09 11:08 ` Zong Li
2020-03-24 7:30 ` [PATCH RFC 4/8] riscv/kaslr: randomize the kernel image offset Zong Li
2020-04-07 5:11 ` Alex Ghiti
2020-04-07 10:53 ` Zong Li
2020-04-09 5:51 ` Alex Ghiti
2020-04-09 10:31 ` Zong Li
2020-04-10 15:58 ` Alex Ghiti
2020-04-11 8:20 ` Zong Li
2020-04-12 6:53 ` Alex Ghiti
2020-04-14 2:46 ` Zong Li
2020-04-14 5:43 ` Alex Ghiti
2020-04-14 7:00 ` Zong Li
2020-03-24 7:30 ` [PATCH RFC 5/8] riscv/kaslr: support sparse memory model Zong Li
2020-03-24 7:30 ` [PATCH RFC 6/8] riscv/kaslr: clear the original kernel image Zong Li
2020-04-07 5:11 ` Alex Ghiti
2020-04-07 11:18 ` Zong Li [this message]
2020-04-09 5:53 ` Alex Ghiti
2020-04-09 8:14 ` Alex Ghiti
2020-03-24 7:30 ` [PATCH RFC 7/8] riscv/kaslr: add cmdline support to disable KASLR Zong Li
2020-03-24 7:30 ` [PATCH RFC 8/8] riscv/kaslr: dump out kernel offset information on panic Zong Li
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CANXhq0p7A2HeyFDRQaw5brzembsFM0-v3kPwQKvFZuyeCm6tsg@mail.gmail.com \
--to=zong.li@sifive.com \
--cc=alex@ghiti.fr \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-riscv@lists.infradead.org \
--cc=palmer@dabbelt.com \
--cc=paul.walmsley@sifive.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).