From: Lai Jiangshan <jiangshanlai@gmail.com>
To: linux-kernel@vger.kernel.org
Cc: Lai Jiangshan <jiangshan.ljs@antgroup.com>,
Hou Wenlong <houwenlong.hwl@antgroup.com>,
Linus Torvalds <torvalds@linux-foundation.org>,
Peter Zijlstra <peterz@infradead.org>,
Sean Christopherson <seanjc@google.com>,
Thomas Gleixner <tglx@linutronix.de>,
Borislav Petkov <bp@alien8.de>, Ingo Molnar <mingo@redhat.com>,
kvm@vger.kernel.org, Paolo Bonzini <pbonzini@redhat.com>,
x86@kernel.org, Kees Cook <keescook@chromium.org>,
Juergen Gross <jgross@suse.com>,
Dave Hansen <dave.hansen@linux.intel.com>,
"H. Peter Anvin" <hpa@zytor.com>
Subject: [RFC PATCH 41/73] KVM: x86/PVM: Allow to load guest TLS in host GDT
Date: Mon, 26 Feb 2024 22:35:58 +0800 [thread overview]
Message-ID: <20240226143630.33643-42-jiangshanlai@gmail.com> (raw)
In-Reply-To: <20240226143630.33643-1-jiangshanlai@gmail.com>
From: Lai Jiangshan <jiangshan.ljs@antgroup.com>
The 32-bit process needs to use TLS in libc, so a hypercall is
introduced to load the guest TLS into the host GDT. The checking of the
guest TLS is the same as tls_desc_okay() in the arch/x86/kernel/tls.c
file.
Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
Signed-off-by: Hou Wenlong <houwenlong.hwl@antgroup.com>
---
arch/x86/kvm/pvm/pvm.c | 81 ++++++++++++++++++++++++++++++++++++++++++
arch/x86/kvm/pvm/pvm.h | 1 +
2 files changed, 82 insertions(+)
diff --git a/arch/x86/kvm/pvm/pvm.c b/arch/x86/kvm/pvm/pvm.c
index ee55e99fb204..e68052f33186 100644
--- a/arch/x86/kvm/pvm/pvm.c
+++ b/arch/x86/kvm/pvm/pvm.c
@@ -281,6 +281,26 @@ static void segments_save_guest_and_switch_to_host(struct vcpu_pvm *pvm)
wrmsrl(MSR_FS_BASE, current->thread.fsbase);
}
+/*
+ * Load guest TLS entries into the GDT.
+ */
+static inline void host_gdt_set_tls(struct vcpu_pvm *pvm)
+{
+ struct desc_struct *gdt = get_current_gdt_rw();
+ unsigned int i;
+
+ for (i = 0; i < GDT_ENTRY_TLS_ENTRIES; i++)
+ gdt[GDT_ENTRY_TLS_MIN + i] = pvm->tls_array[i];
+}
+
+/*
+ * Load current task's TLS into the GDT.
+ */
+static inline void host_gdt_restore_tls(void)
+{
+ native_load_tls(¤t->thread, smp_processor_id());
+}
+
static void pvm_prepare_switch_to_guest(struct kvm_vcpu *vcpu)
{
struct vcpu_pvm *pvm = to_pvm(vcpu);
@@ -304,6 +324,8 @@ static void pvm_prepare_switch_to_guest(struct kvm_vcpu *vcpu)
native_tss_invalidate_io_bitmap();
#endif
+ host_gdt_set_tls(pvm);
+
#ifdef CONFIG_MODIFY_LDT_SYSCALL
/* PVM doesn't support LDT. */
if (unlikely(current->mm->context.ldt))
@@ -334,6 +356,8 @@ static void pvm_prepare_switch_to_host(struct vcpu_pvm *pvm)
kvm_load_ldt(GDT_ENTRY_LDT*8);
#endif
+ host_gdt_restore_tls();
+
segments_save_guest_and_switch_to_host(pvm);
pvm->loaded_cpu_state = 0;
}
@@ -1629,6 +1653,60 @@ static int handle_hc_wrmsr(struct kvm_vcpu *vcpu, u32 index, u64 value)
return 1;
}
+// Check if the tls desc is allowed on the host GDT.
+// The same logic as tls_desc_okay() in arch/x86/kernel/tls.c.
+static bool tls_desc_okay(struct desc_struct *desc)
+{
+ // Only allow present segments.
+ if (!desc->p)
+ return false;
+
+ // Only allow data segments.
+ if (desc->type & (1 << 3))
+ return false;
+
+ // Only allow 32-bit data segments.
+ if (!desc->d)
+ return false;
+
+ return true;
+}
+
+/*
+ * Hypercall: PVM_HC_LOAD_TLS
+ * Load guest TLS desc into host GDT.
+ */
+static int handle_hc_load_tls(struct kvm_vcpu *vcpu, unsigned long tls_desc_0,
+ unsigned long tls_desc_1, unsigned long tls_desc_2)
+{
+ struct vcpu_pvm *pvm = to_pvm(vcpu);
+ unsigned long *tls_array = (unsigned long *)&pvm->tls_array[0];
+ int i;
+
+ tls_array[0] = tls_desc_0;
+ tls_array[1] = tls_desc_1;
+ tls_array[2] = tls_desc_2;
+
+ for (i = 0; i < GDT_ENTRY_TLS_ENTRIES; i++) {
+ if (!tls_desc_okay(&pvm->tls_array[i])) {
+ pvm->tls_array[i] = (struct desc_struct){0};
+ continue;
+ }
+ /* Standarding TLS descs, same as fill_ldt(). */
+ pvm->tls_array[i].type |= 1;
+ pvm->tls_array[i].s = 1;
+ pvm->tls_array[i].dpl = 0x3;
+ pvm->tls_array[i].l = 0;
+ }
+
+ preempt_disable();
+ if (pvm->loaded_cpu_state)
+ host_gdt_set_tls(pvm);
+ preempt_enable();
+
+ return 1;
+}
+
static int handle_kvm_hypercall(struct kvm_vcpu *vcpu)
{
int r;
@@ -1679,6 +1757,8 @@ static int handle_exit_syscall(struct kvm_vcpu *vcpu)
return handle_hc_rdmsr(vcpu, a0);
case PVM_HC_WRMSR:
return handle_hc_wrmsr(vcpu, a0, a1);
+ case PVM_HC_LOAD_TLS:
+ return handle_hc_load_tls(vcpu, a0, a1, a2);
default:
return handle_kvm_hypercall(vcpu);
}
@@ -2296,6 +2376,7 @@ static void pvm_vcpu_reset(struct kvm_vcpu *vcpu, bool init_event)
pvm->hw_ss = __USER_DS;
pvm->int_shadow = 0;
pvm->nmi_mask = false;
+ memset(&pvm->tls_array[0], 0, sizeof(pvm->tls_array));
pvm->msr_vcpu_struct = 0;
pvm->msr_supervisor_rsp = 0;
diff --git a/arch/x86/kvm/pvm/pvm.h b/arch/x86/kvm/pvm/pvm.h
index 31060831e009..f28ab0b48f40 100644
--- a/arch/x86/kvm/pvm/pvm.h
+++ b/arch/x86/kvm/pvm/pvm.h
@@ -98,6 +98,7 @@ struct vcpu_pvm {
struct kvm_segment segments[NR_VCPU_SREG];
struct desc_ptr idt_ptr;
struct desc_ptr gdt_ptr;
+ struct desc_struct tls_array[GDT_ENTRY_TLS_ENTRIES];
};
struct kvm_pvm {
--
2.19.1.6.gb485710b
next prev parent reply other threads:[~2024-02-26 14:36 UTC|newest]
Thread overview: 82+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-02-26 14:35 [RFC PATCH 00/73] KVM: x86/PVM: Introduce a new hypervisor Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 01/73] KVM: Documentation: Add the specification for PVM Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 02/73] x86/ABI/PVM: Add PVM-specific ABI header file Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 03/73] x86/entry: Implement switcher for PVM VM enter/exit Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 04/73] x86/entry: Implement direct switching for the switcher Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 05/73] KVM: x86: Set 'vcpu->arch.exception.injected' as true before vendor callback Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 06/73] KVM: x86: Move VMX interrupt/nmi handling into kvm.ko Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 07/73] KVM: x86/mmu: Adapt shadow MMU for PVM Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 08/73] KVM: x86: Allow hypercall handling to not skip the instruction Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 09/73] KVM: x86: Add PVM virtual MSRs into emulated_msrs_all[] Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 10/73] KVM: x86: Introduce vendor feature to expose vendor-specific CPUID Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 11/73] KVM: x86: Implement gpc refresh for guest usage Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 12/73] KVM: x86: Add NR_VCPU_SREG in SREG enum Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 13/73] KVM: x86/emulator: Reinject #GP if instruction emulation failed for PVM Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 14/73] KVM: x86: Create stubs for PVM module as a new vendor Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 15/73] mm/vmalloc: Add a helper to reserve a contiguous and aligned kernel virtual area Lai Jiangshan
2024-02-27 14:56 ` Christoph Hellwig
2024-02-27 17:07 ` Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 16/73] KVM: x86/PVM: Implement host mmu initialization Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 17/73] KVM: x86/PVM: Implement module initialization related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 18/73] KVM: x86/PVM: Implement VM/VCPU " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 19/73] x86/entry: Export 32-bit ignore syscall entry and __ia32_enabled variable Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 20/73] KVM: x86/PVM: Implement vcpu_load()/vcpu_put() related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 21/73] KVM: x86/PVM: Implement vcpu_run() callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 22/73] KVM: x86/PVM: Handle some VM exits before enable interrupts Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 23/73] KVM: x86/PVM: Handle event handling related MSR read/write operation Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 24/73] KVM: x86/PVM: Introduce PVM mode switching Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 25/73] KVM: x86/PVM: Implement APIC emulation related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 26/73] KVM: x86/PVM: Implement event delivery flags " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 27/73] KVM: x86/PVM: Implement event injection " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 28/73] KVM: x86/PVM: Handle syscall from user mode Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 29/73] KVM: x86/PVM: Implement allowed range checking for #PF Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 30/73] KVM: x86/PVM: Implement segment related callbacks Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 31/73] KVM: x86/PVM: Implement instruction emulation for #UD and #GP Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 32/73] KVM: x86/PVM: Enable guest debugging functions Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 33/73] KVM: x86/PVM: Handle VM-exit due to hardware exceptions Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 34/73] KVM: x86/PVM: Handle ERETU/ERETS synthetic instruction Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 35/73] KVM: x86/PVM: Handle PVM_SYNTHETIC_CPUID " Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 36/73] KVM: x86/PVM: Handle KVM hypercall Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 37/73] KVM: x86/PVM: Use host PCID to reduce guest TLB flushing Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 38/73] KVM: x86/PVM: Handle hypercalls for privilege instruction emulation Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 39/73] KVM: x86/PVM: Handle hypercall for CR3 switching Lai Jiangshan
2024-02-26 14:35 ` [RFC PATCH 40/73] KVM: x86/PVM: Handle hypercall for loading GS selector Lai Jiangshan
2024-02-26 14:35 ` Lai Jiangshan [this message]
2024-02-26 14:35 ` [RFC PATCH 42/73] KVM: x86/PVM: Support for kvm_exit() tracepoint Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 43/73] KVM: x86/PVM: Enable direct switching Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 44/73] KVM: x86/PVM: Implement TSC related callbacks Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 45/73] KVM: x86/PVM: Add dummy PMU " Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 46/73] KVM: x86/PVM: Support for CPUID faulting Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 47/73] KVM: x86/PVM: Handle the left supported MSRs in msrs_to_save_base[] Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 48/73] KVM: x86/PVM: Implement system registers setting callbacks Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 49/73] KVM: x86/PVM: Implement emulation for non-PVM mode Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 50/73] x86/tools/relocs: Cleanup cmdline options Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 51/73] x86/tools/relocs: Append relocations into input file Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 52/73] x86/boot: Allow to do relocation for uncompressed kernel Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 53/73] x86/pvm: Add Kconfig option and the CPU feature bit for PVM guest Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 54/73] x86/pvm: Detect PVM hypervisor support Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 55/73] x86/pvm: Relocate kernel image to specific virtual address range Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 56/73] x86/pvm: Relocate kernel image early in PVH entry Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 57/73] x86/pvm: Make cpu entry area and vmalloc area variable Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 58/73] x86/pvm: Relocate kernel address space layout Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 59/73] x86/pti: Force enabling KPTI for PVM guest Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 60/73] x86/pvm: Add event entry/exit and dispatch code Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 61/73] x86/pvm: Allow to install a system interrupt handler Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 62/73] x86/pvm: Add early kernel event entry and dispatch code Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 63/73] x86/pvm: Add hypercall support Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 64/73] x86/pvm: Enable PVM event delivery Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 65/73] x86/kvm: Patch KVM hypercall as PVM hypercall Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 66/73] x86/pvm: Use new cpu feature to describe XENPV and PVM Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 67/73] x86/pvm: Implement cpu related PVOPS Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 68/73] x86/pvm: Implement irq " Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 69/73] x86/pvm: Implement mmu " Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 70/73] x86/pvm: Don't use SWAPGS for gsbase read/write Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 71/73] x86/pvm: Adapt pushf/popf in this_cpu_cmpxchg16b_emu() Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 72/73] x86/pvm: Use RDTSCP as default in vdso_read_cpunode() Lai Jiangshan
2024-02-26 14:36 ` [RFC PATCH 73/73] x86/pvm: Disable some unsupported syscalls and features Lai Jiangshan
2024-02-26 14:49 ` [RFC PATCH 00/73] KVM: x86/PVM: Introduce a new hypervisor Paolo Bonzini
2024-02-27 17:27 ` Sean Christopherson
2024-02-29 9:33 ` David Woodhouse
2024-03-01 14:00 ` Lai Jiangshan
2024-02-29 14:55 ` Lai Jiangshan
2024-03-06 11:05 ` Like Xu
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=20240226143630.33643-42-jiangshanlai@gmail.com \
--to=jiangshanlai@gmail.com \
--cc=bp@alien8.de \
--cc=dave.hansen@linux.intel.com \
--cc=houwenlong.hwl@antgroup.com \
--cc=hpa@zytor.com \
--cc=jgross@suse.com \
--cc=jiangshan.ljs@antgroup.com \
--cc=keescook@chromium.org \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=pbonzini@redhat.com \
--cc=peterz@infradead.org \
--cc=seanjc@google.com \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.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