[PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions

linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed

* [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions
@ 2020-06-02 18:42 Brendan Shanks
  2020-06-04  0:12 ` Ricardo Neri
  0 siblings, 1 reply; 5+ messages in thread
From: Brendan Shanks @ 2020-06-02 18:42 UTC (permalink / raw)
  To: linux-kernel
  Cc: ricardo.neri-calderon, tglx, mingo, bp, hpa, x86, ebiederm, andi,
	Babu.Moger, Brendan Shanks

Add emulation/spoofing of SLDT and STR for both 32- and 64-bit
processes.

Wine users have found a small number of Windows apps using SLDT that
were crashing when run on UMIP-enabled systems.

Reported-by: Andreas Rammhold <andi@notmuch.email>
Originally-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Signed-off-by: Brendan Shanks <bshanks@codeweavers.com>
---
 arch/x86/kernel/umip.c | 23 ++++++++++++++---------
 1 file changed, 14 insertions(+), 9 deletions(-)

diff --git a/arch/x86/kernel/umip.c b/arch/x86/kernel/umip.c
index 8d5cbe1bbb3b..59dfceac5cc0 100644
--- a/arch/x86/kernel/umip.c
+++ b/arch/x86/kernel/umip.c
@@ -64,6 +64,8 @@
 #define UMIP_DUMMY_GDT_BASE 0xfffffffffffe0000ULL
 #define UMIP_DUMMY_IDT_BASE 0xffffffffffff0000ULL
 
+#define UMIP_DUMMY_TASK_REGISTER_SELECTOR 0x40
+
 /*
  * The SGDT and SIDT instructions store the contents of the global descriptor
  * table and interrupt table registers, respectively. The destination is a
@@ -244,16 +246,24 @@ static int emulate_umip_insn(struct insn *insn, int umip_inst,
 		*data_size += UMIP_GDT_IDT_LIMIT_SIZE;
 		memcpy(data, &dummy_limit, UMIP_GDT_IDT_LIMIT_SIZE);
 
-	} else if (umip_inst == UMIP_INST_SMSW) {
-		unsigned long dummy_value = CR0_STATE;
+	} else if (umip_inst == UMIP_INST_SMSW || umip_inst == UMIP_INST_SLDT ||
+		   umip_inst == UMIP_INST_STR) {
+		unsigned long dummy_value;
+
+		if (umip_inst == UMIP_INST_SMSW)
+			dummy_value = CR0_STATE;
+		else if (umip_inst == UMIP_INST_STR)
+			dummy_value = UMIP_DUMMY_TASK_REGISTER_SELECTOR;
+		else
+			dummy_value = 0;
 
 		/*
-		 * Even though the CR0 register has 4 bytes, the number
+		 * For these 3 instructions, the number
 		 * of bytes to be copied in the result buffer is determined
 		 * by whether the operand is a register or a memory location.
 		 * If operand is a register, return as many bytes as the operand
 		 * size. If operand is memory, return only the two least
-		 * siginificant bytes of CR0.
+		 * siginificant bytes.
 		 */
 		if (X86_MODRM_MOD(insn->modrm.value) == 3)
 			*data_size = insn->opnd_bytes;
@@ -261,7 +271,6 @@ static int emulate_umip_insn(struct insn *insn, int umip_inst,
 			*data_size = 2;
 
 		memcpy(data, &dummy_value, *data_size);
-	/* STR and SLDT  are not emulated */
 	} else {
 		return -EINVAL;
 	}
@@ -383,10 +392,6 @@ bool fixup_umip_exception(struct pt_regs *regs)
 	umip_pr_warn(regs, "%s instruction cannot be used by applications.\n",
 			umip_insns[umip_inst]);
 
-	/* Do not emulate (spoof) SLDT or STR. */
-	if (umip_inst == UMIP_INST_STR || umip_inst == UMIP_INST_SLDT)
-		return false;
-
 	umip_pr_warn(regs, "For now, expensive software emulation returns the result.\n");
 
 	if (emulate_umip_insn(&insn, umip_inst, dummy_data, &dummy_data_size,
-- 
2.26.2


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions
  2020-06-02 18:42 [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions Brendan Shanks
@ 2020-06-04  0:12 ` Ricardo Neri
  2020-06-04  4:39   ` Andy Lutomirski
  0 siblings, 1 reply; 5+ messages in thread
From: Ricardo Neri @ 2020-06-04  0:12 UTC (permalink / raw)
  To: Brendan Shanks
  Cc: linux-kernel, tglx, mingo, bp, hpa, x86, ebiederm, andi, Babu.Moger

On Tue, Jun 02, 2020 at 11:42:12AM -0700, Brendan Shanks wrote:
> Add emulation/spoofing of SLDT and STR for both 32- and 64-bit
> processes.
> 
> Wine users have found a small number of Windows apps using SLDT that
> were crashing when run on UMIP-enabled systems.
> 
> Reported-by: Andreas Rammhold <andi@notmuch.email>
> Originally-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
> Signed-off-by: Brendan Shanks <bshanks@codeweavers.com>
> ---
>  arch/x86/kernel/umip.c | 23 ++++++++++++++---------
>  1 file changed, 14 insertions(+), 9 deletions(-)
> 
> diff --git a/arch/x86/kernel/umip.c b/arch/x86/kernel/umip.c
> index 8d5cbe1bbb3b..59dfceac5cc0 100644
> --- a/arch/x86/kernel/umip.c
> +++ b/arch/x86/kernel/umip.c
> @@ -64,6 +64,8 @@
>  #define UMIP_DUMMY_GDT_BASE 0xfffffffffffe0000ULL
>  #define UMIP_DUMMY_IDT_BASE 0xffffffffffff0000ULL
>  
> +#define UMIP_DUMMY_TASK_REGISTER_SELECTOR 0x40
> +
>  /*
>   * The SGDT and SIDT instructions store the contents of the global descriptor
>   * table and interrupt table registers, respectively. The destination is a
> @@ -244,16 +246,24 @@ static int emulate_umip_insn(struct insn *insn, int umip_inst,
>  		*data_size += UMIP_GDT_IDT_LIMIT_SIZE;
>  		memcpy(data, &dummy_limit, UMIP_GDT_IDT_LIMIT_SIZE);
>  
> -	} else if (umip_inst == UMIP_INST_SMSW) {
> -		unsigned long dummy_value = CR0_STATE;
> +	} else if (umip_inst == UMIP_INST_SMSW || umip_inst == UMIP_INST_SLDT ||
> +		   umip_inst == UMIP_INST_STR) {
> +		unsigned long dummy_value;
> +
> +		if (umip_inst == UMIP_INST_SMSW)
> +			dummy_value = CR0_STATE;
> +		else if (umip_inst == UMIP_INST_STR)
> +			dummy_value = UMIP_DUMMY_TASK_REGISTER_SELECTOR;
> +		else
> +			dummy_value = 0;

Perhaps you can return a non-zero value for SLDT if it has an LDT, as
Andy had suggested. Maybe this can be implemented by looking at
current->mm->context.ldt

I guess the non-zero value can be (GDT_ENTRY_LDT*8).

Thanks and BR,
Ricardo

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions
  2020-06-04  0:12 ` Ricardo Neri
@ 2020-06-04  4:39   ` Andy Lutomirski
       [not found]     ` <E32838F8-665E-488F-96E4-040DD7BDA284@codeweavers.com>
  0 siblings, 1 reply; 5+ messages in thread
From: Andy Lutomirski @ 2020-06-04  4:39 UTC (permalink / raw)
  To: Ricardo Neri
  Cc: Brendan Shanks, LKML, Thomas Gleixner, Ingo Molnar,
	Borislav Petkov, H. Peter Anvin, X86 ML, Eric W. Biederman,
	Andreas Rammhold, Moger, Babu

On Wed, Jun 3, 2020 at 5:12 PM Ricardo Neri
<ricardo.neri-calderon@linux.intel.com> wrote:
>
> On Tue, Jun 02, 2020 at 11:42:12AM -0700, Brendan Shanks wrote:
> > Add emulation/spoofing of SLDT and STR for both 32- and 64-bit
> > processes.
> >
> > Wine users have found a small number of Windows apps using SLDT that
> > were crashing when run on UMIP-enabled systems.
> >
> > Reported-by: Andreas Rammhold <andi@notmuch.email>
> > Originally-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
> > Signed-off-by: Brendan Shanks <bshanks@codeweavers.com>
> > ---
> >  arch/x86/kernel/umip.c | 23 ++++++++++++++---------
> >  1 file changed, 14 insertions(+), 9 deletions(-)
> >
> > diff --git a/arch/x86/kernel/umip.c b/arch/x86/kernel/umip.c
> > index 8d5cbe1bbb3b..59dfceac5cc0 100644
> > --- a/arch/x86/kernel/umip.c
> > +++ b/arch/x86/kernel/umip.c
> > @@ -64,6 +64,8 @@
> >  #define UMIP_DUMMY_GDT_BASE 0xfffffffffffe0000ULL
> >  #define UMIP_DUMMY_IDT_BASE 0xffffffffffff0000ULL
> >
> > +#define UMIP_DUMMY_TASK_REGISTER_SELECTOR 0x40
> > +
> >  /*
> >   * The SGDT and SIDT instructions store the contents of the global descriptor
> >   * table and interrupt table registers, respectively. The destination is a
> > @@ -244,16 +246,24 @@ static int emulate_umip_insn(struct insn *insn, int umip_inst,
> >               *data_size += UMIP_GDT_IDT_LIMIT_SIZE;
> >               memcpy(data, &dummy_limit, UMIP_GDT_IDT_LIMIT_SIZE);
> >
> > -     } else if (umip_inst == UMIP_INST_SMSW) {
> > -             unsigned long dummy_value = CR0_STATE;
> > +     } else if (umip_inst == UMIP_INST_SMSW || umip_inst == UMIP_INST_SLDT ||
> > +                umip_inst == UMIP_INST_STR) {
> > +             unsigned long dummy_value;
> > +
> > +             if (umip_inst == UMIP_INST_SMSW)
> > +                     dummy_value = CR0_STATE;
> > +             else if (umip_inst == UMIP_INST_STR)
> > +                     dummy_value = UMIP_DUMMY_TASK_REGISTER_SELECTOR;
> > +             else
> > +                     dummy_value = 0;
>
> Perhaps you can return a non-zero value for SLDT if it has an LDT, as
> Andy had suggested. Maybe this can be implemented by looking at
> current->mm->context.ldt
>
> I guess the non-zero value can be (GDT_ENTRY_LDT*8).

You could probably even get away with always returning a nonzero
value.  After all, an empty LDT is quite similar to no LDT.

--Andy

^ permalink raw reply	[flat|nested] 5+ messages in thread

[parent not found: <E32838F8-665E-488F-96E4-040DD7BDA284@codeweavers.com>]

* Re: [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions
       [not found]     ` <E32838F8-665E-488F-96E4-040DD7BDA284@codeweavers.com>
@ 2020-06-05 22:42       ` Ricardo Neri
  2020-06-08 18:13         ` Brendan Shanks
  0 siblings, 1 reply; 5+ messages in thread
From: Ricardo Neri @ 2020-06-05 22:42 UTC (permalink / raw)
  To: Brendan Shanks
  Cc: Andy Lutomirski, LKML, Thomas Gleixner, Ingo Molnar,
	Borislav Petkov, H. Peter Anvin, X86 ML, Eric W. Biederman,
	Andreas Rammhold, Moger, Babu

On Fri, Jun 05, 2020 at 11:58:13AM -0700, Brendan Shanks wrote:
> 
> > On Jun 3, 2020, at 9:39 PM, Andy Lutomirski <luto@amacapital.net> wrote:
> > 
> > On Wed, Jun 3, 2020 at 5:12 PM Ricardo Neri
> > <ricardo.neri-calderon@linux.intel.com <mailto:ricardo.neri-calderon@linux.intel.com>> wrote:
> >> 
> >> On Tue, Jun 02, 2020 at 11:42:12AM -0700, Brendan Shanks wrote:
> >>> Add emulation/spoofing of SLDT and STR for both 32- and 64-bit
> >>> processes.
> >>> 
> >>> Wine users have found a small number of Windows apps using SLDT that
> >>> were crashing when run on UMIP-enabled systems.
> >>> 
> >>> Reported-by: Andreas Rammhold <andi@notmuch.email>
> >>> Originally-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
> >>> Signed-off-by: Brendan Shanks <bshanks@codeweavers.com>
> >>> ---
> >>> arch/x86/kernel/umip.c | 23 ++++++++++++++---------
> >>> 1 file changed, 14 insertions(+), 9 deletions(-)
> >>> 
> >>> diff --git a/arch/x86/kernel/umip.c b/arch/x86/kernel/umip.c
> >>> index 8d5cbe1bbb3b..59dfceac5cc0 100644
> >>> --- a/arch/x86/kernel/umip.c
> >>> +++ b/arch/x86/kernel/umip.c
> >>> @@ -64,6 +64,8 @@
> >>> #define UMIP_DUMMY_GDT_BASE 0xfffffffffffe0000ULL
> >>> #define UMIP_DUMMY_IDT_BASE 0xffffffffffff0000ULL
> >>> 
> >>> +#define UMIP_DUMMY_TASK_REGISTER_SELECTOR 0x40
> >>> +
> >>> /*
> >>>  * The SGDT and SIDT instructions store the contents of the global descriptor
> >>>  * table and interrupt table registers, respectively. The destination is a
> >>> @@ -244,16 +246,24 @@ static int emulate_umip_insn(struct insn *insn, int umip_inst,
> >>>              *data_size += UMIP_GDT_IDT_LIMIT_SIZE;
> >>>              memcpy(data, &dummy_limit, UMIP_GDT_IDT_LIMIT_SIZE);
> >>> 
> >>> -     } else if (umip_inst == UMIP_INST_SMSW) {
> >>> -             unsigned long dummy_value = CR0_STATE;
> >>> +     } else if (umip_inst == UMIP_INST_SMSW || umip_inst == UMIP_INST_SLDT ||
> >>> +                umip_inst == UMIP_INST_STR) {
> >>> +             unsigned long dummy_value;
> >>> +
> >>> +             if (umip_inst == UMIP_INST_SMSW)
> >>> +                     dummy_value = CR0_STATE;
> >>> +             else if (umip_inst == UMIP_INST_STR)
> >>> +                     dummy_value = UMIP_DUMMY_TASK_REGISTER_SELECTOR;
> >>> +             else
> >>> +                     dummy_value = 0;
> >> 
> >> Perhaps you can return a non-zero value for SLDT if it has an LDT, as
> >> Andy had suggested. Maybe this can be implemented by looking at
> >> current->mm->context.ldt
> >> 
> >> I guess the non-zero value can be (GDT_ENTRY_LDT*8).
> > 
> > You could probably even get away with always returning a nonzero
> > value.  After all, an empty LDT is quite similar to no LDT.
> 
> 
> Is something like this what you both had in mind?

> I don’t have any software handy to test the LDT-present case though.

Perhaps you can insert a test in the kernel selftest. Something like
this (based on Andreas' test program):

--- a/tools/testing/selftests/x86/ldt_gdt.c
+++ b/tools/testing/selftests/x86/ldt_gdt.c
@@ -220,12 +220,23 @@ static void install_invalid(const struct user_desc *desc, bool oldmode)
 	}
 }

+unsigned long test(void)
+{
+	 unsigned char ldtr[5] = "\xef\xbe\xad\xde";
+	 unsigned long ldt = 0;
+	 asm("sldt %0\n" : "=m" (ldtr));
+	 ldt = *((unsigned long *)&ldtr[0]);
+	 printf ("LDT base: 0x%lx\n", ldt);
+	 return (ldt);
+}
+
 static int safe_modify_ldt(int func, struct user_desc *ptr,
                           unsigned long bytecount)
 {
 	int ret = syscall(SYS_modify_ldt, 0x11, ptr, bytecount);
 	if (ret < -1)
 	errno = -ret;
+	test();
 	return ret;
 }

Thanks and BR,
Ricardo
> 
> 
>                 else if (umip_inst == UMIP_INST_STR)
>                         dummy_value = UMIP_DUMMY_TASK_REGISTER_SELECTOR;
>                 else if (umip_inst == UMIP_INST_SLDT)
>                 {
> #ifdef CONFIG_MODIFY_LDT_SYSCALL
>                         down_read(&current->mm->context.ldt_usr_sem);
>                         if (current->mm->context.ldt)
>                                 dummy_value = GDT_ENTRY_LDT * 8;
>                         else
>                                 dummy_value = 0;
>                         up_read(&current->mm->context.ldt_usr_sem);
> #else
>                         dummy_value = 0;
> #endif
> 

It looks fine to me. Perhaps Andy prefers a simpler, always-non-zero
implementation?

Thanks and BR,
Ricardo

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions
  2020-06-05 22:42       ` Ricardo Neri
@ 2020-06-08 18:13         ` Brendan Shanks
  0 siblings, 0 replies; 5+ messages in thread
From: Brendan Shanks @ 2020-06-08 18:13 UTC (permalink / raw)
  To: Ricardo Neri
  Cc: Andy Lutomirski, LKML, Thomas Gleixner, Ingo Molnar,
	Borislav Petkov, H. Peter Anvin, X86 ML, Eric W. Biederman,
	Andreas Rammhold, Moger, Babu


> On Jun 5, 2020, at 3:42 PM, Ricardo Neri <ricardo.neri-calderon@linux.intel.com> wrote:
> 
> On Fri, Jun 05, 2020 at 11:58:13AM -0700, Brendan Shanks wrote:
>> 
>>> On Jun 3, 2020, at 9:39 PM, Andy Lutomirski <luto@amacapital.net> wrote:
>>> 
>>> On Wed, Jun 3, 2020 at 5:12 PM Ricardo Neri
>>> <ricardo.neri-calderon@linux.intel.com <mailto:ricardo.neri-calderon@linux.intel.com>> wrote:
>>>> 
>>>> On Tue, Jun 02, 2020 at 11:42:12AM -0700, Brendan Shanks wrote:
>>>>> Add emulation/spoofing of SLDT and STR for both 32- and 64-bit
>>>>> processes.
>>>>> 
>>>>> Wine users have found a small number of Windows apps using SLDT that
>>>>> were crashing when run on UMIP-enabled systems.
>>>>> 
>>>>> Reported-by: Andreas Rammhold <andi@notmuch.email>
>>>>> Originally-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
>>>>> Signed-off-by: Brendan Shanks <bshanks@codeweavers.com>
>>>>> ---
>>>>> arch/x86/kernel/umip.c | 23 ++++++++++++++---------
>>>>> 1 file changed, 14 insertions(+), 9 deletions(-)
>>>>> 
>>>>> diff --git a/arch/x86/kernel/umip.c b/arch/x86/kernel/umip.c
>>>>> index 8d5cbe1bbb3b..59dfceac5cc0 100644
>>>>> --- a/arch/x86/kernel/umip.c
>>>>> +++ b/arch/x86/kernel/umip.c
>>>>> @@ -64,6 +64,8 @@
>>>>> #define UMIP_DUMMY_GDT_BASE 0xfffffffffffe0000ULL
>>>>> #define UMIP_DUMMY_IDT_BASE 0xffffffffffff0000ULL
>>>>> 
>>>>> +#define UMIP_DUMMY_TASK_REGISTER_SELECTOR 0x40
>>>>> +
>>>>> /*
>>>>> * The SGDT and SIDT instructions store the contents of the global descriptor
>>>>> * table and interrupt table registers, respectively. The destination is a
>>>>> @@ -244,16 +246,24 @@ static int emulate_umip_insn(struct insn *insn, int umip_inst,
>>>>>             *data_size += UMIP_GDT_IDT_LIMIT_SIZE;
>>>>>             memcpy(data, &dummy_limit, UMIP_GDT_IDT_LIMIT_SIZE);
>>>>> 
>>>>> -     } else if (umip_inst == UMIP_INST_SMSW) {
>>>>> -             unsigned long dummy_value = CR0_STATE;
>>>>> +     } else if (umip_inst == UMIP_INST_SMSW || umip_inst == UMIP_INST_SLDT ||
>>>>> +                umip_inst == UMIP_INST_STR) {
>>>>> +             unsigned long dummy_value;
>>>>> +
>>>>> +             if (umip_inst == UMIP_INST_SMSW)
>>>>> +                     dummy_value = CR0_STATE;
>>>>> +             else if (umip_inst == UMIP_INST_STR)
>>>>> +                     dummy_value = UMIP_DUMMY_TASK_REGISTER_SELECTOR;
>>>>> +             else
>>>>> +                     dummy_value = 0;
>>>> 
>>>> Perhaps you can return a non-zero value for SLDT if it has an LDT, as
>>>> Andy had suggested. Maybe this can be implemented by looking at
>>>> current->mm->context.ldt
>>>> 
>>>> I guess the non-zero value can be (GDT_ENTRY_LDT*8).
>>> 
>>> You could probably even get away with always returning a nonzero
>>> value.  After all, an empty LDT is quite similar to no LDT.
>> 
>> 
>> Is something like this what you both had in mind?
> 
>> I don’t have any software handy to test the LDT-present case though.
> 
> Perhaps you can insert a test in the kernel selftest. Something like
> this (based on Andreas' test program):
> 
> --- a/tools/testing/selftests/x86/ldt_gdt.c
> +++ b/tools/testing/selftests/x86/ldt_gdt.c
> @@ -220,12 +220,23 @@ static void install_invalid(const struct user_desc *desc, bool oldmode)
> 	}
> }
> 
> +unsigned long test(void)
> +{
> +	 unsigned char ldtr[5] = "\xef\xbe\xad\xde";
> +	 unsigned long ldt = 0;
> +	 asm("sldt %0\n" : "=m" (ldtr));
> +	 ldt = *((unsigned long *)&ldtr[0]);
> +	 printf ("LDT base: 0x%lx\n", ldt);
> +	 return (ldt);
> +}
> +
> static int safe_modify_ldt(int func, struct user_desc *ptr,
>                           unsigned long bytecount)
> {
> 	int ret = syscall(SYS_modify_ldt, 0x11, ptr, bytecount);
> 	if (ret < -1)
> 	errno = -ret;
> +	test();
> 	return ret;
> }

Thanks for the tip, I gave that a try and got the same results under UMIP and a non-UMIP system. I’ll send this next version of the patch out.

Brendan Shanks
CodeWeavers


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2020-06-08 18:13 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-06-02 18:42 [PATCH] x86/umip: Add emulation/spoofing for SLDT and STR instructions Brendan Shanks
2020-06-04  0:12 ` Ricardo Neri
2020-06-04  4:39   ` Andy Lutomirski
     [not found]     ` <E32838F8-665E-488F-96E4-040DD7BDA284@codeweavers.com>
2020-06-05 22:42       ` Ricardo Neri
2020-06-08 18:13         ` Brendan Shanks

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).