mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Peter Zijlstra <peterz@infradead.org>
To: Tianyu Lan <ltykernel@gmail.com>
Cc: kys@microsoft.com, haiyangz@microsoft.com, wei.liu@kernel.org,
	decui@microsoft.com, tglx@linutronix.de, mingo@redhat.com,
	bp@alien8.de, dave.hansen@linux.intel.com, x86@kernel.org,
	hpa@zytor.com, daniel.lezcano@linaro.org, arnd@arndb.de,
	michael.h.kelley@microsoft.com, Tianyu Lan <tiala@microsoft.com>,
	linux-arch@vger.kernel.org, linux-hyperv@vger.kernel.org,
	linux-kernel@vger.kernel.org, vkuznets@redhat.com
Subject: Re: [EXTERNAL] Re: [PATCH 5/9] x86/hyperv: Use vmmcall to implement Hyper-V hypercall in sev-snp enlightened guest
Date: Tue, 27 Jun 2023 13:50:02 +0200	[thread overview]
Message-ID: <20230627115002.GW83892@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <d06bb33e-047f-c849-de6a-246bc361c7af@gmail.com>

On Tue, Jun 27, 2023 at 06:57:28PM +0800, Tianyu Lan wrote:

> > "There is no x86 SEV SNP feature(X86_FEATURE_SEV_SNP) flag

I'm sure we can arrange such a feature if we need it, this isn't rocket
science. Boris?

> > support so far and hardware provides MSR_AMD64_SEV register
> > to check SEV-SNP capability with MSR_AMD64_SEV_ENABLED bit
> > ALTERNATIVE can't work without SEV-SNP x86 feature flag."
> > There is no cpuid leaf bit to check AMD SEV-SNP feature.
> > 
> > After some Hyper-V doesn't provides SEV and SEV-ES guest before and so
> > may reuse X86_FEATURE_SEV and X86_FEATURE_SEV_ES flag as alternative
> > feature check for Hyper-V SEV-SNP guest. Will refresh patch.
> > 
> 
> Hi Peter:
>      I tried using alternative for "vmmcall" and CALL_NOSPEC in a single
> Inline assembly. The output is different in the SEV-SNP mode. When SEV-
> SNP is enabled, thunk_target is not required. While it's necessary in
> the non SEV-SNP mode. Do you have any idea how to differentiate outputs in
> the single Inline assembly which just like alternative works for
> assembler template.

This seems to work; it's a bit magical for having a nested ALTERNATIVE
but the output seems correct (the outer alternative comes last in
.altinstructions and thus has precedence). Sure the [thunk_target] input
goes unsed in one of the alteratives, but who cares.


static inline u64 hv_do_hypercall(u64 control, void *input, void *output)
{
	u64 input_address = input ? virt_to_phys(input) : 0;
	u64 output_address = output ? virt_to_phys(output) : 0;
	u64 hv_status;

#ifdef CONFIG_X86_64
	if (!hv_hypercall_pg)
		return U64_MAX;

#if 0
	if (hv_isolation_type_en_snp()) {
		__asm__ __volatile__("mov %4, %%r8\n"
				     "vmmcall"
				     : "=a" (hv_status), ASM_CALL_CONSTRAINT,
				       "+c" (control), "+d" (input_address)
				     :  "r" (output_address)
				     : "cc", "memory", "r8", "r9", "r10", "r11");
	} else {
		__asm__ __volatile__("mov %4, %%r8\n"
				     CALL_NOSPEC
				     : "=a" (hv_status), ASM_CALL_CONSTRAINT,
				       "+c" (control), "+d" (input_address)
				     :  "r" (output_address),
					THUNK_TARGET(hv_hypercall_pg)
				     : "cc", "memory", "r8", "r9", "r10", "r11");
	}
#endif

	asm volatile("mov %[output], %%r8\n"
		     ALTERNATIVE(CALL_NOSPEC, "vmmcall", X86_FEATURE_SEV_ES)
		     : "=a" (hv_status), ASM_CALL_CONSTRAINT,
		       "+c" (control), "+d" (input_address)
		     : [output] "r" (output_address),
		       THUNK_TARGET(hv_hypercall_pg)
		     : "cc", "memory", "r8", "r9", "r10", "r11");

#else
	u32 input_address_hi = upper_32_bits(input_address);
	u32 input_address_lo = lower_32_bits(input_address);
	u32 output_address_hi = upper_32_bits(output_address);
	u32 output_address_lo = lower_32_bits(output_address);

	if (!hv_hypercall_pg)
		return U64_MAX;

	__asm__ __volatile__(CALL_NOSPEC
			     : "=A" (hv_status),
			       "+c" (input_address_lo), ASM_CALL_CONSTRAINT
			     : "A" (control),
			       "b" (input_address_hi),
			       "D"(output_address_hi), "S"(output_address_lo),
			       THUNK_TARGET(hv_hypercall_pg)
			     : "cc", "memory");
#endif /* !x86_64 */
	return hv_status;
}

(in actual fact x86_64-defconfig + kvm_guest.config + HYPERV)

$ ./scripts/objdump-func defconfig-build/arch/x86/hyperv/mmu.o hv_do_hypercall
0000 0000000000000cd0 <hv_do_hypercall.constprop.0>:
0000      cd0:  48 89 f9                mov    %rdi,%rcx
0003      cd3:  31 d2                   xor    %edx,%edx
0005      cd5:  48 85 f6                test   %rsi,%rsi
0008      cd8:  74 1b                   je     cf5 <hv_do_hypercall.constprop.0+0x25>
000a      cda:  b8 00 00 00 80          mov    $0x80000000,%eax
000f      cdf:  48 01 c6                add    %rax,%rsi
0012      ce2:  72 38                   jb     d1c <hv_do_hypercall.constprop.0+0x4c>
0014      ce4:  48 c7 c2 00 00 00 80    mov    $0xffffffff80000000,%rdx
001b      ceb:  48 2b 15 00 00 00 00    sub    0x0(%rip),%rdx        # cf2 <hv_do_hypercall.constprop.0+0x22>   cee: R_X86_64_PC32      page_offset_base-0x4
0022      cf2:  48 01 f2                add    %rsi,%rdx
0025      cf5:  48 8b 35 00 00 00 00    mov    0x0(%rip),%rsi        # cfc <hv_do_hypercall.constprop.0+0x2c>   cf8: R_X86_64_PC32      hv_hypercall_pg-0x4
002c      cfc:  48 85 f6                test   %rsi,%rsi
002f      cff:  74 0f                   je     d10 <hv_do_hypercall.constprop.0+0x40>
0031      d01:  31 c0                   xor    %eax,%eax
0033      d03:  49 89 c0                mov    %rax,%r8
0036      d06:  ff d6                   call   *%rsi
0038      d08:  90                      nop
0039      d09:  90                      nop
003a      d0a:  90                      nop
003b      d0b:  e9 00 00 00 00          jmp    d10 <hv_do_hypercall.constprop.0+0x40>   d0c: R_X86_64_PLT32     __x86_return_thunk-0x4
0040      d10:  48 c7 c0 ff ff ff ff    mov    $0xffffffffffffffff,%rax
0047      d17:  e9 00 00 00 00          jmp    d1c <hv_do_hypercall.constprop.0+0x4c>   d18: R_X86_64_PLT32     __x86_return_thunk-0x4
004c      d1c:  48 8b 15 00 00 00 00    mov    0x0(%rip),%rdx        # d23 <hv_do_hypercall.constprop.0+0x53>   d1f: R_X86_64_PC32      phys_base-0x4
0053      d23:  eb cd                   jmp    cf2 <hv_do_hypercall.constprop.0+0x22>

$ objdump -wdr -j .altinstr_replacement defconfig-build/arch/x86/hyperv/mmu.o
0000000000000000 <.altinstr_replacement>:
0:   f3 48 0f b8 c7          popcnt %rdi,%rax
5:   e8 00 00 00 00          call   a <.altinstr_replacement+0xa>    6: R_X86_64_PLT32       __x86_indirect_thunk_rsi-0x4
a:   0f ae e8                lfence
d:   ff d6                   call   *%rsi
f:   0f 01 d9                vmmcall

$ ./readelf-section.sh defconfig-build/arch/x86/hyperv/mmu.o altinstructions
Relocation section '.rela.altinstructions' at offset 0x5420 contains 8 entries:
Offset             Info             Type               Symbol's Value  Symbol's Name + Addend
0000000000000000  0000000200000002 R_X86_64_PC32          0000000000000000 .text + 1e3
0000000000000004  0000000700000002 R_X86_64_PC32          0000000000000000 .altinstr_replacement + 0
000000000000000e  0000000200000002 R_X86_64_PC32          0000000000000000 .text + d06
0000000000000012  0000000700000002 R_X86_64_PC32          0000000000000000 .altinstr_replacement + 5
000000000000001c  0000000200000002 R_X86_64_PC32          0000000000000000 .text + d06
0000000000000020  0000000700000002 R_X86_64_PC32          0000000000000000 .altinstr_replacement + a
000000000000002a  0000000200000002 R_X86_64_PC32          0000000000000000 .text + d06
000000000000002e  0000000700000002 R_X86_64_PC32          0000000000000000 .altinstr_replacement + f


  reply	other threads:[~2023-06-27 11:50 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-06-01 15:16 [PATCH 0/9] x86/hyperv: Add AMD sev-snp enlightened guest support on hyperv Tianyu Lan
2023-06-01 15:16 ` [PATCH 1/9] x86/hyperv: Add sev-snp enlightened guest static key Tianyu Lan
2023-06-05 12:09   ` Vitaly Kuznetsov
2023-06-06 13:43     ` Tianyu Lan
2023-07-18  5:52     ` Tianyu Lan
2023-06-08 12:56   ` Michael Kelley (LINUX)
2023-06-08 13:17     ` Tianyu Lan
2023-06-01 15:16 ` [PATCH 2/9] x86/hyperv: Set Virtual Trust Level in VMBus init message Tianyu Lan
2023-06-08 13:06   ` Michael Kelley (LINUX)
2023-06-08 13:21     ` Tianyu Lan
2023-06-01 15:16 ` [PATCH 3/9] x86/hyperv: Mark Hyper-V vp assist page unencrypted in SEV-SNP enlightened guest Tianyu Lan
2023-06-05 12:13   ` Vitaly Kuznetsov
2023-06-06 15:22     ` Tianyu Lan
2023-06-06 15:49       ` Vitaly Kuznetsov
2023-06-08 13:25         ` Michael Kelley (LINUX)
2023-06-08 13:44           ` Vitaly Kuznetsov
2023-06-01 15:16 ` [PATCH 4/9] drivers: hv: Mark shared pages " Tianyu Lan
2023-06-05 12:54   ` Vitaly Kuznetsov
2023-06-07  8:16     ` Tianyu Lan
2023-06-08  8:54       ` Vitaly Kuznetsov
2023-06-08 14:21   ` Michael Kelley (LINUX)
2023-06-01 15:16 ` [PATCH 5/9] x86/hyperv: Use vmmcall to implement Hyper-V hypercall in sev-snp " Tianyu Lan
2023-06-05 13:00   ` Vitaly Kuznetsov
2023-06-08 13:21   ` Peter Zijlstra
2023-06-08 15:15     ` [EXTERNAL] " Tianyu Lan
2023-06-27 10:57       ` Tianyu Lan
2023-06-27 11:50         ` Peter Zijlstra [this message]
2023-06-27 12:05           ` Borislav Petkov
2023-06-27 13:38             ` Peter Zijlstra
2023-06-28 10:53               ` Peter Zijlstra
2023-06-01 15:16 ` [PATCH 6/9] clocksource: hyper-v: Mark hyperv tsc page unencrypted " Tianyu Lan
2023-06-01 15:16 ` [PATCH 7/9] x86/hyperv: Initialize cpu and memory for SEV-SNP " Tianyu Lan
2023-06-08 13:51   ` Michael Kelley (LINUX)
2023-06-09  9:56     ` Jeremi Piotrowski
2023-06-08 14:09   ` Michael Kelley (LINUX)
2023-06-08 15:18     ` Tianyu Lan
2023-06-01 15:16 ` [PATCH 8/9] x86/hyperv: Add smp support for SEV-SNP guest Tianyu Lan
2023-06-01 15:16 ` [PATCH 9/9] x86/hyperv: Add hyperv-specific handling for VMMCALL under SEV-ES Tianyu Lan

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=20230627115002.GW83892@hirez.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=arnd@arndb.de \
    --cc=bp@alien8.de \
    --cc=daniel.lezcano@linaro.org \
    --cc=dave.hansen@linux.intel.com \
    --cc=decui@microsoft.com \
    --cc=haiyangz@microsoft.com \
    --cc=hpa@zytor.com \
    --cc=kys@microsoft.com \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-hyperv@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=ltykernel@gmail.com \
    --cc=michael.h.kelley@microsoft.com \
    --cc=mingo@redhat.com \
    --cc=tglx@linutronix.de \
    --cc=tiala@microsoft.com \
    --cc=vkuznets@redhat.com \
    --cc=wei.liu@kernel.org \
    --cc=x86@kernel.org \
    /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

Powered by JetHome