From: Andrew Jones <drjones@redhat.com>
To: Alexandru Elisei <alexandru.elisei@arm.com>
Cc: kvm@vger.kernel.org, nikos.nikoleris@arm.com,
andre.przywara@arm.com, eric.auger@redhat.com
Subject: Re: [PATCH kvm-unit-tests 1/8] arm/arm64: Reorganize cstart assembler
Date: Wed, 14 Apr 2021 10:59:21 +0200 [thread overview]
Message-ID: <20210414085921.lazllz24o3eqts52@kamzik.brq.redhat.com> (raw)
In-Reply-To: <2b647637-d307-5256-beab-c58728f60e9b@arm.com>
On Tue, Apr 13, 2021 at 05:34:24PM +0100, Alexandru Elisei wrote:
> Hi Drew,
>
> On 4/7/21 7:59 PM, Andrew Jones wrote:
> > Move secondary_entry helper functions out of .init and into .text,
> > since secondary_entry isn't run at "init" time.
>
> The tests aren't loaded using the loader, so as far as I can tell the reason for
> having an .init section is to make sure the code from the start label is put at
> offset 0 in the test binary. As long as the start label is kept at the beginning
> of the .init section, and the loader script places the section first, I don't see
> any issues with this change.
>
> The only hypothetical problem that I can think of is that the code from .init
> calls code from .text, and if the text section grows very large we might end up
> with a PC offset larger than what can be encoded in the BL instruction. That's
> unlikely to happen (the offset is 16MB for arm and 64MB for arm64), and the .init
> code already calls other functions (like setup) which are in .text, so we would
> have this problem regardless of this change. And the compiler will emit an error
> if that happens.
>
> >
> > Signed-off-by: Andrew Jones <drjones@redhat.com>
> > ---
> > arm/cstart.S | 62 +++++++++++++++++++++++++++-----------------------
> > arm/cstart64.S | 22 +++++++++++-------
> > 2 files changed, 48 insertions(+), 36 deletions(-)
> >
> > diff --git a/arm/cstart.S b/arm/cstart.S
> > index d88a98362940..653ab1e8a141 100644
> > --- a/arm/cstart.S
> > +++ b/arm/cstart.S
> > @@ -96,32 +96,7 @@ start:
> > bl exit
> > b halt
> >
> > -
> > -.macro set_mode_stack mode, stack
> > - add \stack, #S_FRAME_SIZE
> > - msr cpsr_c, #(\mode | PSR_I_BIT | PSR_F_BIT)
> > - isb
> > - mov sp, \stack
> > -.endm
> > -
> > -exceptions_init:
> > - mrc p15, 0, r2, c1, c0, 0 @ read SCTLR
> > - bic r2, #CR_V @ SCTLR.V := 0
> > - mcr p15, 0, r2, c1, c0, 0 @ write SCTLR
> > - ldr r2, =vector_table
> > - mcr p15, 0, r2, c12, c0, 0 @ write VBAR
> > -
> > - mrs r2, cpsr
> > -
> > - /* first frame reserved for svc mode */
> > - set_mode_stack UND_MODE, r0
> > - set_mode_stack ABT_MODE, r0
> > - set_mode_stack IRQ_MODE, r0
> > - set_mode_stack FIQ_MODE, r0
> > -
> > - msr cpsr_cxsf, r2 @ back to svc mode
> > - isb
> > - mov pc, lr
> > +.text
>
> Hm... now we've moved enable_vfp from .init to .text, and enable_vfp *is* called
> from .init code, which doesn't fully match up with the commit message. Is the
> actual reason for this change that the linker script for EFI will discard the
> .init section? Maybe it's worth mentioning that in the commit message, because it
> will explain this change better.
Right, the .init section may not exist when linking with other linker
scripts. I'll make the commit message more clear.
> Or is it to align arm with arm64, where only
> start is in the .init section?
>
> >
> > enable_vfp:
> > /* Enable full access to CP10 and CP11: */
> > @@ -133,8 +108,6 @@ enable_vfp:
> > vmsr fpexc, r0
> > mov pc, lr
> >
> > -.text
> > -
> > .global get_mmu_off
> > get_mmu_off:
> > ldr r0, =auxinfo
> > @@ -235,6 +208,39 @@ asm_mmu_disable:
> >
> > mov pc, lr
> >
> > +/*
> > + * Vectors
> > + */
> > +
> > +.macro set_mode_stack mode, stack
> > + add \stack, #S_FRAME_SIZE
> > + msr cpsr_c, #(\mode | PSR_I_BIT | PSR_F_BIT)
> > + isb
> > + mov sp, \stack
> > +.endm
> > +
> > +exceptions_init:
> > + mrc p15, 0, r2, c1, c0, 0 @ read SCTLR
> > + bic r2, #CR_V @ SCTLR.V := 0
> > + mcr p15, 0, r2, c1, c0, 0 @ write SCTLR
> > + ldr r2, =vector_table
> > + mcr p15, 0, r2, c12, c0, 0 @ write VBAR
> > +
> > + mrs r2, cpsr
> > +
> > + /*
> > + * Input r0 is the stack top, which is the exception stacks base
> > + * The first frame is reserved for svc mode
> > + */
> > + set_mode_stack UND_MODE, r0
> > + set_mode_stack ABT_MODE, r0
> > + set_mode_stack IRQ_MODE, r0
> > + set_mode_stack FIQ_MODE, r0
> > +
> > + msr cpsr_cxsf, r2 @ back to svc mode
> > + isb
> > + mov pc, lr
> > +
> > /*
> > * Vector stubs
> > * Simplified version of the Linux kernel implementation
> > diff --git a/arm/cstart64.S b/arm/cstart64.S
> > index 0a85338bcdae..d39cf4dfb99c 100644
> > --- a/arm/cstart64.S
> > +++ b/arm/cstart64.S
> > @@ -89,10 +89,12 @@ start:
> > msr cpacr_el1, x4
> >
> > /* set up exception handling */
> > + mov x4, x0 // x0 is the addr of the dtb
>
> I suppose changing exceptions_init to use x0 as a scratch register instead of x4
> makes some sense if you look at it from the perspective of it being called from
> secondary_entry, where all the functions use x0 as a scratch register. But it's
> still called from start, where using x4 as a scratch register is preferred because
> of the kernel boot protocol (x0-x3 are reserved).
>
> Is there an actual bug that this is supposed to fix (I looked for it and couldn't
> figure it out) or is it just a cosmetic change?
Now that exceptions_init isn't a private function of start (actually it
hasn't been in a long time, considering secondary_entry calls it) I would
like it to better conform to calling conventions. I guess I should have
used x19 here instead of x4 to be 100% correct. Or, would you rather I
just continue using x4 in exceptions_init in order to avoid the
save/restore?
Thanks,
drew
next prev parent reply other threads:[~2021-04-14 9:00 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-04-07 18:59 [PATCH kvm-unit-tests 0/8] arm/arm64: Prepare for target-efi Andrew Jones
2021-04-07 18:59 ` [PATCH kvm-unit-tests 1/8] arm/arm64: Reorganize cstart assembler Andrew Jones
2021-04-09 17:18 ` Nikos Nikoleris
2021-04-09 17:28 ` Andrew Jones
2021-04-13 16:34 ` Alexandru Elisei
2021-04-14 8:59 ` Andrew Jones [this message]
2021-04-14 15:15 ` Alexandru Elisei
2021-04-15 13:03 ` Andrew Jones
2021-04-07 18:59 ` [PATCH kvm-unit-tests 2/8] arm/arm64: Move setup_vm into setup Andrew Jones
2021-04-09 17:24 ` Nikos Nikoleris
2021-04-14 15:19 ` Alexandru Elisei
2021-04-07 18:59 ` [PATCH kvm-unit-tests 3/8] pci-testdev: ioremap regions Andrew Jones
2021-04-13 14:12 ` Nikos Nikoleris
2021-04-07 18:59 ` [PATCH kvm-unit-tests 4/8] arm/arm64: mmu: Stop mapping an assumed IO region Andrew Jones
2021-04-13 14:06 ` Nikos Nikoleris
2021-04-14 15:42 ` Alexandru Elisei
2021-04-15 13:09 ` Andrew Jones
2021-04-07 18:59 ` [PATCH kvm-unit-tests 5/8] arm/arm64: mmu: Remove memory layout assumptions Andrew Jones
2021-04-13 14:27 ` Nikos Nikoleris
2021-04-15 15:48 ` Alexandru Elisei
2021-04-15 17:11 ` Andrew Jones
2021-04-19 15:09 ` Alexandru Elisei
2021-04-07 18:59 ` [PATCH kvm-unit-tests 6/8] arm/arm64: setup: Consolidate " Andrew Jones
2021-04-13 16:41 ` Nikos Nikoleris
2021-04-14 9:03 ` Andrew Jones
2021-04-15 16:59 ` Alexandru Elisei
2021-04-15 17:25 ` Andrew Jones
2021-04-19 15:56 ` Alexandru Elisei
2021-04-19 15:59 ` Alexandru Elisei
2021-04-19 17:53 ` Andrew Jones
2021-04-07 18:59 ` [PATCH kvm-unit-tests 7/8] chr-testdev: Silently fail init Andrew Jones
2021-04-13 16:42 ` Nikos Nikoleris
2021-04-15 17:03 ` Alexandru Elisei
2021-04-07 18:59 ` [PATCH kvm-unit-tests 8/8] arm/arm64: psci: don't assume method is hvc Andrew Jones
2021-04-09 17:46 ` Nikos Nikoleris
2021-04-14 9:06 ` Andrew Jones
2021-04-19 16:33 ` Alexandru Elisei
2021-04-19 18:13 ` Andrew Jones
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=20210414085921.lazllz24o3eqts52@kamzik.brq.redhat.com \
--to=drjones@redhat.com \
--cc=alexandru.elisei@arm.com \
--cc=andre.przywara@arm.com \
--cc=eric.auger@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=nikos.nikoleris@arm.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).