From: Euan Harris <euan.harris@citrix.com>
To: xen-devel@lists.xenproject.org
Cc: andrew.cooper3@citrix.com, kevin.tian@intel.com,
Euan Harris <euan.harris@citrix.com>,
jun.nakajima@intel.com, jbeulich@suse.com
Subject: [PATCH 2/9] x86/vvmx: Unify operands in struct vmx_inst_decoded
Date: Thu, 26 Oct 2017 18:03:12 +0100 [thread overview]
Message-ID: <1509037399-48926-3-git-send-email-euan.harris@citrix.com> (raw)
In-Reply-To: <1509037399-48926-1-git-send-email-euan.harris@citrix.com>
This change prepares the way for abstracting the code which reads
memory and register operands into a generic operand_read() function in
a future patch.
Operand 1 may either be a memory location or a register, depending on
the instruction being handled. Operand 2 may only be a register, but
it is now represented in the same way as operand 1 and so has a type
field. This will always equal VMX_INST_MEMREG_TYPE_REG.
References to the old structure map to the new one as follows:
mem -> op[0].mem
len -> op[0].len
reg1 -> op[0].reg
reg2 -> op[1].reg
References to type always map to op[0].type in this patch because
previously operand 2 could only be a register and so type was never
checked when accessing it.
Signed-off-by: Euan Harris <euan.harris@citrix.com>
---
xen/arch/x86/hvm/vmx/vvmx.c | 52 ++++++++++++++++++++++++---------------------
1 file changed, 28 insertions(+), 24 deletions(-)
diff --git a/xen/arch/x86/hvm/vmx/vvmx.c b/xen/arch/x86/hvm/vmx/vvmx.c
index 9d87786894..20e5e29031 100644
--- a/xen/arch/x86/hvm/vmx/vvmx.c
+++ b/xen/arch/x86/hvm/vmx/vvmx.c
@@ -195,16 +195,16 @@ bool_t nvmx_ept_enabled(struct vcpu *v)
struct vmx_inst_decoded {
#define VMX_INST_MEMREG_TYPE_MEMORY 0
#define VMX_INST_MEMREG_TYPE_REG 1
- int type;
- union {
- struct {
- unsigned long mem;
- unsigned int len;
+ struct vmx_inst_op {
+ int type;
+ union {
+ struct {
+ unsigned long mem;
+ unsigned int len;
+ };
+ unsigned int reg_idx;
};
- unsigned int reg1;
- };
-
- unsigned int reg2;
+ } op[2];
};
enum vmx_ops_result {
@@ -437,16 +437,16 @@ static int decode_vmx_inst(struct cpu_user_regs *regs,
info.word = offset;
if ( info.fields.memreg ) {
- decode->type = VMX_INST_MEMREG_TYPE_REG;
- decode->reg1 = info.fields.reg1;
+ decode->op[0].type = VMX_INST_MEMREG_TYPE_REG;
+ decode->op[0].reg_idx = info.fields.reg1;
if ( poperandS != NULL )
- *poperandS = reg_read(regs, decode->reg1);
+ *poperandS = reg_read(regs, decode->op[0].reg_idx);
}
else
{
bool mode_64bit = (vmx_guest_x86_mode(v) == 8);
- decode->type = VMX_INST_MEMREG_TYPE_MEMORY;
+ decode->op[0].type = VMX_INST_MEMREG_TYPE_MEMORY;
if ( info.fields.segment > x86_seg_gs )
goto gp_fault;
@@ -486,11 +486,13 @@ static int decode_vmx_inst(struct cpu_user_regs *regs,
if ( rc != HVMTRANS_okay )
return X86EMUL_EXCEPTION;
}
- decode->mem = base;
- decode->len = size;
+
+ decode->op[0].mem = base;
+ decode->op[0].len = size;
}
- decode->reg2 = info.fields.reg2;
+ decode->op[1].type = VMX_INST_MEMREG_TYPE_REG;
+ decode->op[1].reg_idx = info.fields.reg2;
return X86EMUL_OKAY;
@@ -1770,7 +1772,8 @@ int nvmx_handle_vmptrst(struct cpu_user_regs *regs)
gpa = nvcpu->nv_vvmcxaddr;
- rc = hvm_copy_to_guest_linear(decode.mem, &gpa, decode.len, 0, &pfinfo);
+ rc = hvm_copy_to_guest_linear(decode.op[0].mem, &gpa,
+ decode.op[0].len, 0, &pfinfo);
if ( rc == HVMTRANS_bad_linear_to_gfn )
hvm_inject_page_fault(pfinfo.ec, pfinfo.linear);
if ( rc != HVMTRANS_okay )
@@ -1850,23 +1853,24 @@ int nvmx_handle_vmread(struct cpu_user_regs *regs)
return X86EMUL_OKAY;
}
- rc = get_vvmcs_safe(v, reg_read(regs, decode.reg2), &value);
+ rc = get_vvmcs_safe(v, reg_read(regs, decode.op[1].reg_idx), &value);
if ( rc != VMX_INSN_SUCCEED )
{
vmfail(regs, rc);
return X86EMUL_OKAY;
}
- switch ( decode.type ) {
+ switch ( decode.op[0].type ) {
case VMX_INST_MEMREG_TYPE_MEMORY:
- rc = hvm_copy_to_guest_linear(decode.mem, &value, decode.len, 0, &pfinfo);
+ rc = hvm_copy_to_guest_linear(decode.op[0].mem, &value,
+ decode.op[0].len, 0, &pfinfo);
if ( rc == HVMTRANS_bad_linear_to_gfn )
hvm_inject_page_fault(pfinfo.ec, pfinfo.linear);
if ( rc != HVMTRANS_okay )
return X86EMUL_EXCEPTION;
break;
case VMX_INST_MEMREG_TYPE_REG:
- reg_write(regs, decode.reg1, value);
+ reg_write(regs, decode.op[0].reg_idx, value);
break;
}
@@ -1893,7 +1897,7 @@ int nvmx_handle_vmwrite(struct cpu_user_regs *regs)
return X86EMUL_OKAY;
}
- vmcs_encoding = reg_read(regs, decode.reg2);
+ vmcs_encoding = reg_read(regs, decode.op[1].reg_idx);
err = set_vvmcs_safe(v, vmcs_encoding, operand);
if ( err != VMX_INSN_SUCCEED )
{
@@ -1931,7 +1935,7 @@ int nvmx_handle_invept(struct cpu_user_regs *regs)
if ( (ret = decode_vmx_inst(regs, &decode, &eptp, 0)) != X86EMUL_OKAY )
return ret;
- switch ( reg_read(regs, decode.reg2) )
+ switch ( reg_read(regs, decode.op[1].reg_idx) )
{
case INVEPT_SINGLE_CONTEXT:
{
@@ -1959,7 +1963,7 @@ int nvmx_handle_invvpid(struct cpu_user_regs *regs)
if ( (ret = decode_vmx_inst(regs, &decode, &vpid, 0)) != X86EMUL_OKAY )
return ret;
- switch ( reg_read(regs, decode.reg2) )
+ switch ( reg_read(regs, decode.op[1].reg_idx) )
{
/* Just invalidate all tlb entries for all types! */
case INVVPID_INDIVIDUAL_ADDR:
--
2.13.6
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
next prev parent reply other threads:[~2017-10-26 17:03 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-10-26 17:03 [PATCH 0/9] x86/vvmx: Read instruction operands correctly on VM exit Euan Harris
2017-10-26 17:03 ` [PATCH 1/9] x86/vvmx: Remove enum vmx_regs_enc Euan Harris
2017-10-26 17:03 ` Euan Harris [this message]
2017-10-26 17:03 ` [PATCH 3/9] x86/vvmx: Extract operand reading logic into operand_read() Euan Harris
2017-11-27 17:01 ` Jan Beulich
2017-11-27 18:08 ` Andrew Cooper
2017-11-28 8:40 ` Jan Beulich
2017-11-28 8:14 ` Jan Beulich
2017-10-26 17:03 ` [PATCH 4/9] x86/vvmx: Remove unnecessary VMX operand reads Euan Harris
2017-11-02 7:07 ` Tian, Kevin
2017-11-27 17:03 ` Jan Beulich
2017-10-26 17:03 ` [PATCH 5/9] x86/vvmx: Replace direct calls to reg_read() with operand_read() Euan Harris
2017-10-26 17:03 ` [PATCH 6/9] x86/vvmx: Remove operand reading from decode_vmx_inst() Euan Harris
2017-10-26 17:03 ` [PATCH 7/9] x86/vvmx: Use correct sizes when reading operands Euan Harris
2017-11-28 8:32 ` Jan Beulich
2017-12-01 18:45 ` Andrew Cooper
2017-12-04 8:55 ` Jan Beulich
2017-10-26 17:03 ` [PATCH 8/9] x86/hvm: Add hvm_copy_{to, from}_guest_virt() helpers Euan Harris
2017-11-28 8:38 ` Jan Beulich
2017-10-26 17:03 ` [PATCH 9/9] x86/vvmx: Use hvm_copy_{to, from}_guest_virt() to read operands Euan Harris
2017-11-28 8:49 ` Jan Beulich
2017-10-26 17:59 ` [PATCH 0/9] x86/vvmx: Read instruction operands correctly on VM exit Andrew Cooper
2017-11-02 7:23 ` Tian, Kevin
2017-11-02 18:39 ` Andrew Cooper
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=1509037399-48926-3-git-send-email-euan.harris@citrix.com \
--to=euan.harris@citrix.com \
--cc=andrew.cooper3@citrix.com \
--cc=jbeulich@suse.com \
--cc=jun.nakajima@intel.com \
--cc=kevin.tian@intel.com \
--cc=xen-devel@lists.xenproject.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;
as well as URLs for NNTP newsgroup(s).