diff mbox series

[v2,RESEND,4/4] MIPS: emulate CPUCFG instruction on older Loongson64 cores

Message ID 20200503105012.43246-1-git@xen0n.name (mailing list archive)
State Rejected
Headers show
Series None | expand

Commit Message

WANG Xuerui May 3, 2020, 10:50 a.m. UTC
CPUCFG is the instruction for querying processor characteristics on
newer Loongson processors, much like CPUID of x86. Since the instruction
is supposedly designed to provide a unified way to do feature detection
(without having to, for example, parse /proc/cpuinfo which is too
heavyweight), it is important to provide compatibility for older cores
without native support. Fortunately, most of the fields can be
synthesized without changes to semantics. Performance is not really big
a concern, because feature detection logic is not expected to be
invoked very often in typical userland applications.

The instruction can't be emulated on LOONGSON_2EF cores, according to
FlyGoat's experiments. Because the LWC2 opcode is assigned to other
valid instructions on 2E and 2F, no RI exception is raised for us to
intercept. So compatibility is only extended back furthest to
Loongson-3A1000. Loongson-2K is covered too, as it is basically a remix
of various blocks from the 3A/3B models from a kernel perspective.

This is lightly based on Loongson's work on their Linux 3.10 fork, for
being the authority on the right feature flags to fill in, where things
aren't otherwise discoverable.

Signed-off-by: WANG Xuerui <git@xen0n.name>
Reviewed-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Tested-by: Jiaxun Yang <jiaxun.yang@flygoat.com> # Loongson-2K, 3B1500
Cc: Huacai Chen <chenhc@lemote.com>
Cc: Jiaxun Yang <jiaxun.yang@flygoat.com>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
---

Resend because I forgot to include Jiaxun's review tags. Sorry for the
noise.

 arch/mips/Kconfig                             |  11 ++
 arch/mips/include/asm/cpu-info.h              |   9 ++
 .../include/asm/mach-loongson64/cpucfg-emul.h |  67 +++++++++
 arch/mips/kernel/cpu-probe.c                  | 129 ++++++++++++++++++
 arch/mips/kernel/traps.c                      |  45 ++++++
 arch/mips/loongson64/Makefile                 |   1 +
 arch/mips/loongson64/cpucfg-emul.c            |  80 +++++++++++
 7 files changed, 342 insertions(+)
 create mode 100644 arch/mips/include/asm/mach-loongson64/cpucfg-emul.h
 create mode 100644 arch/mips/loongson64/cpucfg-emul.c

Comments

Jiaxun Yang May 3, 2020, 3:50 p.m. UTC | #1
于 2020年5月3日 GMT+08:00 下午6:50:13, WANG Xuerui <git@xen0n.name> 写到:
>CPUCFG is the instruction for querying processor characteristics on
>newer Loongson processors, much like CPUID of x86. Since the instruction
>is supposedly designed to provide a unified way to do feature detection
>(without having to, for example, parse /proc/cpuinfo which is too
>heavyweight), it is important to provide compatibility for older cores
>without native support. Fortunately, most of the fields can be
>synthesized without changes to semantics. Performance is not really big
>a concern, because feature detection logic is not expected to be
>invoked very often in typical userland applications.
>
>The instruction can't be emulated on LOONGSON_2EF cores, according to
>FlyGoat's experiments. Because the LWC2 opcode is assigned to other
>valid instructions on 2E and 2F, no RI exception is raised for us to
>intercept. So compatibility is only extended back furthest to
>Loongson-3A1000. Loongson-2K is covered too, as it is basically a remix
>of various blocks from the 3A/3B models from a kernel perspective.
>
>This is lightly based on Loongson's work on their Linux 3.10 fork, for
>being the authority on the right feature flags to fill in, where things
>aren't otherwise discoverable.
>
>Signed-off-by: WANG Xuerui <git@xen0n.name>
>Reviewed-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
>Tested-by: Jiaxun Yang <jiaxun.yang@flygoat.com> # Loongson-2K, 3B1500
>Cc: Huacai Chen <chenhc@lemote.com>
>Cc: Jiaxun Yang <jiaxun.yang@flygoat.com>
>Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
>---

Some random thoughts on that:

While cpucfg instruction would not be available on other MIPS
processors, and given that most distros are more willing to provide
general MIPS version instead of Loongson specified version, we'd better
provide user space programs a way to probe emulated/physical cpucfg support.

Looks like elf_hwcap would be a option? I've already added some flags about
Loongson's extension and now an extra HWCAP_LOONGSON_CPUCFG can cover everything.

>
>Resend because I forgot to include Jiaxun's review tags. Sorry for the
>noise.
>
> arch/mips/Kconfig                             |  11 ++
> arch/mips/include/asm/cpu-info.h              |   9 ++
> .../include/asm/mach-loongson64/cpucfg-emul.h |  67 +++++++++
> arch/mips/kernel/cpu-probe.c                  | 129 ++++++++++++++++++
> arch/mips/kernel/traps.c                      |  45 ++++++
> arch/mips/loongson64/Makefile                 |   1 +
> arch/mips/loongson64/cpucfg-emul.c            |  80 +++++++++++
> 7 files changed, 342 insertions(+)
> create mode 100644 arch/mips/include/asm/mach-loongson64/cpucfg-emul.h
> create mode 100644 arch/mips/loongson64/cpucfg-emul.c
WANG Xuerui May 4, 2020, 5:25 a.m. UTC | #2
On 5/3/20 11:50 PM, Jiaxun Yang wrote:

> Some random thoughts on that:
>
> While cpucfg instruction would not be available on other MIPS
> processors, and given that most distros are more willing to provide
> general MIPS version instead of Loongson specified version, we'd better
> provide user space programs a way to probe emulated/physical cpucfg support.
>
> Looks like elf_hwcap would be a option? I've already added some flags about
> Loongson's extension and now an extra HWCAP_LOONGSON_CPUCFG can cover everything.

All emulated outputs from the older models have the LCSRP bit clear in 
this patch. As CPUCFG is part of the Loongson CSR ASE, this is the 
reliable way to distinguish between emulated and physical CPUCFG support.

However, because detecting presence of CPUCFG itself still requires one 
to handle SIGILL, exposing an hwcap might be an option. So we'll have 
the following mapping:

- no HWCAP_LOONGSON_CPUCFG -- other MIPS or Loongson-2EF or Loongson-32
- CPUCFG.0x2.LCSRP == 0 -- older Loongson-64 models
- otherwise, Loongson-3A R4 or later

Seems some of the previously defined hwcaps will become unnecessary if 
we make this change. But they're already part of kernel ABI so we have 
to keep them, fortunately we still have plenty of bits available. At 
least future additions of Loongson features wouldn't go through hwcap.

I think the hwcap addition can be added in another follow-up series if 
the idea gets consensus; let's keep this series focused.
Jiaxun Yang May 4, 2020, 6:58 a.m. UTC | #3
于 2020年5月4日 GMT+08:00 下午1:25:00, WANG Xuerui <kernel@xen0n.name> 写到:
>On 5/3/20 11:50 PM, Jiaxun Yang wrote:
>
>> Some random thoughts on that:
>>
>> While cpucfg instruction would not be available on other MIPS
>> processors, and given that most distros are more willing to provide
>> general MIPS version instead of Loongson specified version, we'd better
>> provide user space programs a way to probe emulated/physical cpucfg support.
>>
>> Looks like elf_hwcap would be a option? I've already added some flags about
>> Loongson's extension and now an extra HWCAP_LOONGSON_CPUCFG can cover everything.
>
>All emulated outputs from the older models have the LCSRP bit clear in 
>this patch. As CPUCFG is part of the Loongson CSR ASE, this is the 
>reliable way to distinguish between emulated and physical CPUCFG support.
>
>However, because detecting presence of CPUCFG itself still requires one 
>to handle SIGILL, exposing an hwcap might be an option. So we'll have 
>the following mapping:
>
>- no HWCAP_LOONGSON_CPUCFG -- other MIPS or Loongson-2EF or Loongson-32
>- CPUCFG.0x2.LCSRP == 0 -- older Loongson-64 models
>- otherwise, Loongson-3A R4 or later
>
>Seems some of the previously defined hwcaps will become unnecessary if 
>we make this change. But they're already part of kernel ABI so we have 
>to keep them, fortunately we still have plenty of bits available. At 
>least future additions of Loongson features wouldn't go through hwcap.
>
>I think the hwcap addition can be added in another follow-up series if 
>the idea gets consensus; let's keep this series focused.

Yeah that's exactly my point.

I'm looking forward for getting these stuff merged.

Thanks.

>
Thomas Bogendoerfer May 17, 2020, 8:37 a.m. UTC | #4
On Sun, May 03, 2020 at 06:50:13PM +0800, WANG Xuerui wrote:
> +#endif
> +
> +#endif /* _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_ */
> diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
> index ca2e6f1af4fe..907e31ff562f 100644
> --- a/arch/mips/kernel/cpu-probe.c
> +++ b/arch/mips/kernel/cpu-probe.c
> @@ -28,6 +28,8 @@
>  #include <asm/spram.h>
>  #include <linux/uaccess.h>
>  
> +#include <asm/mach-loongson64/cpucfg-emul.h>
> +

this doesn't fly:

In file included from /local/tbogendoerfer/korg/linux/arch/mips/kernel/cpu-probe.c:31:0:
/local/tbogendoerfer/korg/linux/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h:7:27: fatal error: loongson_regs.h: No such file or directory
 #include <loongson_regs.h>
                           ^
compilation terminated.

Is there a chance to put this code in a loongsoon specific file ?

Thomas.
WANG Xuerui May 17, 2020, 11:39 a.m. UTC | #5
On 5/17/20 4:37 PM, Thomas Bogendoerfer wrote:

> On Sun, May 03, 2020 at 06:50:13PM +0800, WANG Xuerui wrote:
>> +#endif
>> +
>> +#endif /* _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_ */
>> diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
>> index ca2e6f1af4fe..907e31ff562f 100644
>> --- a/arch/mips/kernel/cpu-probe.c
>> +++ b/arch/mips/kernel/cpu-probe.c
>> @@ -28,6 +28,8 @@
>>   #include <asm/spram.h>
>>   #include <linux/uaccess.h>
>>   
>> +#include <asm/mach-loongson64/cpucfg-emul.h>
>> +
> this doesn't fly:
>
> In file included from /local/tbogendoerfer/korg/linux/arch/mips/kernel/cpu-probe.c:31:0:
> /local/tbogendoerfer/korg/linux/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h:7:27: fatal error: loongson_regs.h: No such file or directory
>   #include <loongson_regs.h>
>                             ^
> compilation terminated.
>
> Is there a chance to put this code in a loongsoon specific file ?

Oops... I'll rebase tonight to fix this. Might be caused by the build 
system changes merged in the meantime.

As for the logic separation, I'm 100% in agreement with this, but I 
don't know of any way to invoke mach-specific bits from inside 
cpu-probe. Plus all the PRId matching logic have to be duplicated if the 
code is to be moved to the loongson64/ sub-directory. I didn't find any 
other platform that does this, so I just extended the probing code.

Other way is to treat the whole CPUCFG thing just like the r2-to-r6-emul 
thing, which means moving all the logic to kernel/ sub-directory. But 
I'd rather not do this, for reasons explained in the cover letter.

>
> Thomas.
>
Thomas Bogendoerfer May 17, 2020, 3:17 p.m. UTC | #6
On Sun, May 17, 2020 at 07:39:44PM +0800, WANG Xuerui wrote:
> On 5/17/20 4:37 PM, Thomas Bogendoerfer wrote:
> 
> >On Sun, May 03, 2020 at 06:50:13PM +0800, WANG Xuerui wrote:
> >>+#endif
> >>+
> >>+#endif /* _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_ */
> >>diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
> >>index ca2e6f1af4fe..907e31ff562f 100644
> >>--- a/arch/mips/kernel/cpu-probe.c
> >>+++ b/arch/mips/kernel/cpu-probe.c
> >>@@ -28,6 +28,8 @@
> >>  #include <asm/spram.h>
> >>  #include <linux/uaccess.h>
> >>+#include <asm/mach-loongson64/cpucfg-emul.h>
> >>+
> >this doesn't fly:
> >
> >In file included from /local/tbogendoerfer/korg/linux/arch/mips/kernel/cpu-probe.c:31:0:
> >/local/tbogendoerfer/korg/linux/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h:7:27: fatal error: loongson_regs.h: No such file or directory
> >  #include <loongson_regs.h>
> >                            ^
> >compilation terminated.
> >
> >Is there a chance to put this code in a loongsoon specific file ?
> 
> Oops... I'll rebase tonight to fix this. Might be caused by the build system
> changes merged in the meantime.

just to make it clear, the failing config is a RALINK config, but looking
at the failure probaly everything but loongson64 fails.

> As for the logic separation, I'm 100% in agreement with this, but I don't
> know of any way to invoke mach-specific bits from inside cpu-probe.

implement your code in a new/fitting file, provide a header file,
which has prototypes for this functions if CONFIG_xxx option is enabled
or empty stubs, if not. Then call these functions from cpu-probe.c.

Thomas.
WANG Xuerui May 18, 2020, 2:44 a.m. UTC | #7
On 2020/5/17 23:17, Thomas Bogendoerfer wrote:

> On Sun, May 17, 2020 at 07:39:44PM +0800, WANG Xuerui wrote:
>> On 5/17/20 4:37 PM, Thomas Bogendoerfer wrote:
>>
>>> On Sun, May 03, 2020 at 06:50:13PM +0800, WANG Xuerui wrote:
>>>> +#endif
>>>> +
>>>> +#endif /* _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_ */
>>>> diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
>>>> index ca2e6f1af4fe..907e31ff562f 100644
>>>> --- a/arch/mips/kernel/cpu-probe.c
>>>> +++ b/arch/mips/kernel/cpu-probe.c
>>>> @@ -28,6 +28,8 @@
>>>>   #include <asm/spram.h>
>>>>   #include <linux/uaccess.h>
>>>> +#include <asm/mach-loongson64/cpucfg-emul.h>
>>>> +
>>> this doesn't fly:
>>>
>>> In file included from /local/tbogendoerfer/korg/linux/arch/mips/kernel/cpu-probe.c:31:0:
>>> /local/tbogendoerfer/korg/linux/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h:7:27: fatal error: loongson_regs.h: No such file or directory
>>>   #include <loongson_regs.h>
>>>                             ^
>>> compilation terminated.
>>>
>>> Is there a chance to put this code in a loongsoon specific file ?
>> Oops... I'll rebase tonight to fix this. Might be caused by the build system
>> changes merged in the meantime.
> just to make it clear, the failing config is a RALINK config, but looking
> at the failure probaly everything but loongson64 fails.
Thanks for the clarification; I've found out the mistake last night. The 
include of loongson_regs.h needs to go inside the config guard to 
prevent inclusion on other platforms where the include path is not set 
up for it.
>> As for the logic separation, I'm 100% in agreement with this, but I don't
>> know of any way to invoke mach-specific bits from inside cpu-probe.
> implement your code in a new/fitting file, provide a header file,
> which has prototypes for this functions if CONFIG_xxx option is enabled
> or empty stubs, if not. Then call these functions from cpu-probe.c.

Sure. I'll move these into cpucfg-emul.c to minimize intrusion to 
cpu-probe.c.

I'll send v3 later today (in my timezone) after I go home.

And thanks for your instruction!

> Thomas.
>
WANG Xuerui May 19, 2020, 2:33 p.m. UTC | #8
On 5/18/20 10:44 AM, WANG Xuerui wrote:

>
> Sure. I'll move these into cpucfg-emul.c to minimize intrusion to 
> cpu-probe.c.
>
> I'll send v3 later today (in my timezone) after I go home.
>
Sorry, but my development machine is experiencing some connectivity 
problems, plus this week is rather busier than normal.

v3 probably won't be ready until maybe Thursday. Will try my best to 
submit before the 5.8 merge window though...
diff mbox series

Patch

diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index 9f15539a6342..2ab189001917 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -1508,6 +1508,17 @@  config CPU_LOONGSON3_WORKAROUNDS
 
 	  If unsure, please say Y.
 
+config CPU_LOONGSON3_CPUCFG_EMULATION
+	bool "Emulate the CPUCFG instruction on older cores"
+	default y
+	depends on CPU_LOONGSON64
+	help
+	  Loongson-3A R4 and newer have the CPUCFG instruction available for
+	  userland to query CPU capabilities, much like CPUID on x86. This
+	  option provides emulation of the instruction on older cores.
+
+	  If unsure, please say Y.
+
 config CPU_LOONGSON2E
 	bool "Loongson 2E"
 	depends on SYS_HAS_CPU_LOONGSON2E
diff --git a/arch/mips/include/asm/cpu-info.h b/arch/mips/include/asm/cpu-info.h
index ed7ffe4e63a3..43c238567a80 100644
--- a/arch/mips/include/asm/cpu-info.h
+++ b/arch/mips/include/asm/cpu-info.h
@@ -105,6 +105,15 @@  struct cpuinfo_mips {
 	unsigned int		gtoffset_mask;
 	unsigned int		guestid_mask;
 	unsigned int		guestid_cache;
+
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+	/* CPUCFG data for this CPU, synthesized at probe time.
+	 *
+	 * CPUCFG select 0 is PRId, others are unimplemented for now. So the
+	 * only stored values are for CPUCFG selects 1-3 inclusive.
+	 */
+	u32 loongson3_cpucfg_data[3];
+#endif
 } __attribute__((aligned(SMP_CACHE_BYTES)));
 
 extern struct cpuinfo_mips cpu_data[];
diff --git a/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h b/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h
new file mode 100644
index 000000000000..b9f6b2aa98f9
--- /dev/null
+++ b/arch/mips/include/asm/mach-loongson64/cpucfg-emul.h
@@ -0,0 +1,67 @@ 
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_
+#define _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_
+
+#include <asm/cpu-info.h>
+
+#include <loongson_regs.h>
+
+#define LOONGSON_FPREV_MASK 0x7
+
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+
+/* Finalize synthesis of CPUCFG data by patching the partially filled data
+ * with dynamically detected CPU characteristics. This keeps the amount of
+ * hard-coded logic at a minimum.
+ */
+void loongson3_cpucfg_finish_synthesis(struct cpuinfo_mips *c);
+
+static inline u32 loongson3_cpucfg_read_synthesized(
+	struct cpuinfo_mips *c,
+	__u64 sel)
+{
+	switch (sel) {
+	case LOONGSON_CFG0:
+		return c->processor_id;
+	case LOONGSON_CFG1:
+	case LOONGSON_CFG2:
+	case LOONGSON_CFG3:
+		return c->loongson3_cpucfg_data[sel - 1];
+	case LOONGSON_CFG4:
+	case LOONGSON_CFG5:
+		/* CPUCFG selects 4 and 5 are related to the processor clock.
+		 * Unimplemented for now.
+		 */
+		return 0;
+	case LOONGSON_CFG6:
+		/* CPUCFG select 6 is for the undocumented Safe Extension. */
+		return 0;
+	case LOONGSON_CFG7:
+		/* CPUCFG select 7 is for the virtualization extension.
+		 * We don't know if the two currently known features are
+		 * supported on older cores according to the public
+		 * documentation, so leave this at zero.
+		 */
+		return 0;
+	}
+
+	/*
+	 * Return 0 for unrecognized CPUCFG selects, which is real hardware
+	 * behavior observed on Loongson 3A R4.
+	 */
+	return 0;
+}
+#else
+static void loongson3_cpucfg_finish_synthesis(struct cpuinfo_mips *c)
+{
+}
+
+static inline u32 loongson3_cpucfg_read_synthesized(
+	struct cpuinfo_mips *c,
+	__u64 sel)
+{
+	return 0;
+}
+#endif
+
+#endif /* _ASM_MACH_LOONGSON64_CPUCFG_EMUL_H_ */
diff --git a/arch/mips/kernel/cpu-probe.c b/arch/mips/kernel/cpu-probe.c
index ca2e6f1af4fe..907e31ff562f 100644
--- a/arch/mips/kernel/cpu-probe.c
+++ b/arch/mips/kernel/cpu-probe.c
@@ -28,6 +28,8 @@ 
 #include <asm/spram.h>
 #include <linux/uaccess.h>
 
+#include <asm/mach-loongson64/cpucfg-emul.h>
+
 /* Hardware capabilities */
 unsigned int elf_hwcap __read_mostly;
 EXPORT_SYMBOL_GPL(elf_hwcap);
@@ -1580,6 +1582,24 @@  static inline void cpu_probe_legacy(struct cpuinfo_mips *c, unsigned int cpu)
 			set_isa(c, MIPS_CPU_ISA_M64R1);
 			c->ases |= (MIPS_ASE_LOONGSON_MMI | MIPS_ASE_LOONGSON_CAM |
 				MIPS_ASE_LOONGSON_EXT);
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+			/* Add CPUCFG features non-discoverable otherwise. */
+			c->loongson3_cpucfg_data[0] |= (LOONGSON_CFG1_LSLDR0 |
+				LOONGSON_CFG1_LSSYNCI | LOONGSON_CFG1_LSUCA |
+				LOONGSON_CFG1_LLSYNC | LOONGSON_CFG1_TGTSYNC);
+			c->loongson3_cpucfg_data[1] |= (LOONGSON_CFG2_LBT1 |
+				LOONGSON_CFG2_LPMP | LOONGSON_CFG2_LPM_REV1);
+			c->loongson3_cpucfg_data[2] |= (
+				LOONGSON_CFG3_LCAM_REV1 |
+				LOONGSON_CFG3_LCAMNUM_REV1 |
+				LOONGSON_CFG3_LCAMKW_REV1 |
+				LOONGSON_CFG3_LCAMVW_REV1);
+
+			/* This feature is set by firmware, but all known
+			 * Loongson-3A Legacy systems are configured this way.
+			 */
+			c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_CDMAP;
+#endif
 			break;
 		case PRID_REV_LOONGSON3B_R1:
 		case PRID_REV_LOONGSON3B_R2:
@@ -1589,6 +1609,24 @@  static inline void cpu_probe_legacy(struct cpuinfo_mips *c, unsigned int cpu)
 			set_isa(c, MIPS_CPU_ISA_M64R1);
 			c->ases |= (MIPS_ASE_LOONGSON_MMI | MIPS_ASE_LOONGSON_CAM |
 				MIPS_ASE_LOONGSON_EXT);
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+			/* Add CPUCFG features non-discoverable otherwise. */
+			c->loongson3_cpucfg_data[0] |= (LOONGSON_CFG1_LSLDR0 |
+				LOONGSON_CFG1_LSSYNCI | LOONGSON_CFG1_LSUCA |
+				LOONGSON_CFG1_LLSYNC | LOONGSON_CFG1_TGTSYNC);
+			c->loongson3_cpucfg_data[1] |= (LOONGSON_CFG2_LBT1 |
+				LOONGSON_CFG2_LPMP | LOONGSON_CFG2_LPM_REV1);
+			c->loongson3_cpucfg_data[2] |= (
+				LOONGSON_CFG3_LCAM_REV1 |
+				LOONGSON_CFG3_LCAMNUM_REV1 |
+				LOONGSON_CFG3_LCAMKW_REV1 |
+				LOONGSON_CFG3_LCAMVW_REV1);
+
+			/* This feature is set by firmware, but all known
+			 * Loongson-3B Legacy systems are configured this way.
+			 */
+			c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_CDMAP;
+#endif
 			break;
 		}
 
@@ -1957,9 +1995,57 @@  static inline void decode_cpucfg(struct cpuinfo_mips *c)
 		c->ases |= MIPS_ASE_LOONGSON_CAM;
 }
 
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+static inline int cpu_has_uca(void)
+{
+	u32 diag = read_c0_diag();
+	u32 new_diag;
+
+	if (diag & LOONGSON_DIAG_UCAC)
+		/* UCA is already enabled. */
+		return 1;
+
+	/* See if UCAC bit can be flipped on. This should be safe. */
+	new_diag = diag | LOONGSON_DIAG_UCAC;
+	write_c0_diag(new_diag);
+	new_diag = read_c0_diag();
+	write_c0_diag(diag);
+
+	return (new_diag & LOONGSON_DIAG_UCAC) != 0;
+}
+
+static inline void probe_uca(struct cpuinfo_mips *c)
+{
+	if (cpu_has_uca())
+		c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_LSUCA;
+}
+
+static inline void decode_loongson_config6(struct cpuinfo_mips *c)
+{
+	u32 config6 = read_c0_config6();
+
+	if (config6 & MIPS_CONF6_LOONGSON_STFILL)
+		c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_SFBP;
+	if (config6 & MIPS_CONF6_LOONGSON_LLEXC)
+		c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_LLEXC;
+	if (config6 & MIPS_CONF6_LOONGSON_SCRAND)
+		c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_SCRAND;
+}
+#else
+static inline void probe_uca(struct cpuinfo_mips *c)
+{
+}
+
+static inline void decode_loongson_config6(struct cpuinfo_mips *c)
+{
+}
+#endif /* CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION */
+
 static inline void cpu_probe_loongson(struct cpuinfo_mips *c, unsigned int cpu)
 {
 	decode_configs(c);
+	decode_loongson_config6(c);
+	probe_uca(c);
 
 	switch (c->processor_id & PRID_IMP_MASK) {
 	case PRID_IMP_LOONGSON_64R: /* Loongson-64 Reduced */
@@ -1977,6 +2063,21 @@  static inline void cpu_probe_loongson(struct cpuinfo_mips *c, unsigned int cpu)
 		c->writecombine = _CACHE_UNCACHED_ACCELERATED;
 		c->ases |= (MIPS_ASE_LOONGSON_MMI | MIPS_ASE_LOONGSON_EXT |
 				MIPS_ASE_LOONGSON_EXT2);
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+		/* Add CPUCFG features non-discoverable otherwise. */
+		c->loongson3_cpucfg_data[0] |= (LOONGSON_CFG1_LSLDR0 |
+			LOONGSON_CFG1_LSSYNCI | LOONGSON_CFG1_LLSYNC |
+			LOONGSON_CFG1_TGTSYNC);
+		c->loongson3_cpucfg_data[1] |= (LOONGSON_CFG2_LBT1 |
+			LOONGSON_CFG2_LBT2 | LOONGSON_CFG2_LPMP |
+			LOONGSON_CFG2_LPM_REV2);
+		c->loongson3_cpucfg_data[2] = 0;
+
+		/* This feature is set by firmware, but all known Loongson-2K
+		 * systems are configured this way.
+		 */
+		c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_CDMAP;
+#endif
 		break;
 	case PRID_IMP_LOONGSON_64C:  /* Loongson-3 Classic */
 		switch (c->processor_id & PRID_REV_MASK) {
@@ -2007,6 +2108,26 @@  static inline void cpu_probe_loongson(struct cpuinfo_mips *c, unsigned int cpu)
 		c->writecombine = _CACHE_UNCACHED_ACCELERATED;
 		c->ases |= (MIPS_ASE_LOONGSON_MMI | MIPS_ASE_LOONGSON_CAM |
 			MIPS_ASE_LOONGSON_EXT | MIPS_ASE_LOONGSON_EXT2);
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+		/* Add CPUCFG features non-discoverable otherwise. */
+		c->loongson3_cpucfg_data[0] |= (LOONGSON_CFG1_CNT64 |
+			LOONGSON_CFG1_LSLDR0 | LOONGSON_CFG1_LSPREF |
+			LOONGSON_CFG1_LSPREFX | LOONGSON_CFG1_LSSYNCI |
+			LOONGSON_CFG1_LLSYNC | LOONGSON_CFG1_TGTSYNC);
+		c->loongson3_cpucfg_data[1] |= (LOONGSON_CFG2_LBT1 |
+			LOONGSON_CFG2_LBT2 | LOONGSON_CFG2_LBTMMU |
+			LOONGSON_CFG2_LPMP | LOONGSON_CFG2_LPM_REV1 |
+			LOONGSON_CFG2_LVZ_REV1);
+		c->loongson3_cpucfg_data[2] |= (LOONGSON_CFG3_LCAM_REV1 |
+			LOONGSON_CFG3_LCAMNUM_REV1 |
+			LOONGSON_CFG3_LCAMKW_REV1 |
+			LOONGSON_CFG3_LCAMVW_REV1);
+
+		/* This feature is set by firmware, but all known Loongson-3
+		 * systems are configured this way.
+		 */
+		c->loongson3_cpucfg_data[0] |= LOONGSON_CFG1_CDMAP;
+#endif
 		break;
 	case PRID_IMP_LOONGSON_64G:
 		c->cputype = CPU_LOONGSON64;
@@ -2200,6 +2321,12 @@  void cpu_probe(void)
 	c->fpu_csr31	= FPU_CSR_RN;
 	c->fpu_msk31	= FPU_CSR_RSVD | FPU_CSR_ABS2008 | FPU_CSR_NAN2008;
 
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+	c->loongson3_cpucfg_data[0] = 0;
+	c->loongson3_cpucfg_data[1] = 0;
+	c->loongson3_cpucfg_data[2] = 0;
+#endif
+
 	c->processor_id = read_c0_prid();
 	switch (c->processor_id & PRID_COMP_MASK) {
 	case PRID_COMP_LEGACY:
@@ -2333,6 +2460,8 @@  void cpu_probe(void)
 	if (cpu_has_vz)
 		cpu_probe_vz(c);
 
+	loongson3_cpucfg_finish_synthesis(c);
+
 	cpu_probe_vmbits(c);
 
 #ifdef CONFIG_64BIT
diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index 31968cbd6464..768790917724 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -71,6 +71,8 @@ 
 #include <asm/tlbex.h>
 #include <asm/uasm.h>
 
+#include <asm/mach-loongson64/cpucfg-emul.h>
+
 extern void check_wait(void);
 extern asmlinkage void rollback_handle_int(void);
 extern asmlinkage void handle_int(void);
@@ -693,6 +695,44 @@  static int simulate_sync(struct pt_regs *regs, unsigned int opcode)
 	return -1;			/* Must be something else ... */
 }
 
+/*
+ * Loongson-3 CSR instructions emulation
+ */
+
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+
+#define LWC2             0xc8000000
+#define RS               BASE
+#define CSR_OPCODE2      0x00000118
+#define CSR_OPCODE2_MASK 0x000007ff
+#define CSR_FUNC_MASK    RT
+#define CSR_FUNC_CPUCFG  0x8
+
+static int simulate_loongson3_cpucfg(struct pt_regs *regs,
+				     unsigned int opcode)
+{
+	int op = opcode & OPCODE;
+	int op2 = opcode & CSR_OPCODE2_MASK;
+	int csr_func = (opcode & CSR_FUNC_MASK) >> 16;
+
+	if (op == LWC2 && op2 == CSR_OPCODE2 && csr_func == CSR_FUNC_CPUCFG) {
+		int rd = (opcode & RD) >> 11;
+		int rs = (opcode & RS) >> 21;
+		__u64 sel = regs->regs[rs];
+
+		perf_sw_event(PERF_COUNT_SW_EMULATION_FAULTS, 1, regs, 0);
+
+		regs->regs[rd] = loongson3_cpucfg_read_synthesized(
+			&current_cpu_data, sel);
+
+		return 0;
+	}
+
+	/* Not ours.  */
+	return -1;
+}
+#endif /* CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION */
+
 asmlinkage void do_ov(struct pt_regs *regs)
 {
 	enum ctx_state prev_state;
@@ -1166,6 +1206,11 @@  asmlinkage void do_ri(struct pt_regs *regs)
 
 		if (status < 0)
 			status = simulate_fp(regs, opcode, old_epc, old31);
+
+#ifdef CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION
+		if (status < 0)
+			status = simulate_loongson3_cpucfg(regs, opcode);
+#endif
 	} else if (cpu_has_mmips) {
 		unsigned short mmop[2] = { 0 };
 
diff --git a/arch/mips/loongson64/Makefile b/arch/mips/loongson64/Makefile
index 6f3c2b47f66f..61f6add20530 100644
--- a/arch/mips/loongson64/Makefile
+++ b/arch/mips/loongson64/Makefile
@@ -10,3 +10,4 @@  obj-$(CONFIG_NUMA)	+= numa.o
 obj-$(CONFIG_RS780_HPET) += hpet.o
 obj-$(CONFIG_PCI) += pci.o
 obj-$(CONFIG_SUSPEND) += pm.o
+obj-$(CONFIG_CPU_LOONGSON3_CPUCFG_EMULATION) += cpucfg-emul.o
diff --git a/arch/mips/loongson64/cpucfg-emul.c b/arch/mips/loongson64/cpucfg-emul.c
new file mode 100644
index 000000000000..a30f82b55c9e
--- /dev/null
+++ b/arch/mips/loongson64/cpucfg-emul.c
@@ -0,0 +1,80 @@ 
+// SPDX-License-Identifier: GPL-2.0
+
+#include <linux/smp.h>
+#include <asm/cpu.h>
+#include <asm/cpu-info.h>
+
+#include <loongson_regs.h>
+#include <cpucfg-emul.h>
+
+static u32 get_loongson_fprev(struct cpuinfo_mips *c)
+{
+	return (c->fpu_id & LOONGSON_FPREV_MASK) << LOONGSON_CFG1_FPREV_OFFSET;
+}
+
+static void patch_cpucfg_sel1(struct cpuinfo_mips *c)
+{
+	u64 ases = c->ases;
+	u64 options = c->options;
+	u32 data = c->loongson3_cpucfg_data[0];
+
+	if (options & MIPS_CPU_FPU) {
+		data |= LOONGSON_CFG1_FP;
+		data |= get_loongson_fprev(c);
+	}
+	if (ases & MIPS_ASE_LOONGSON_MMI)
+		data |= LOONGSON_CFG1_MMI;
+	if (ases & MIPS_ASE_MSA)
+		data |= LOONGSON_CFG1_MSA1;
+
+	c->loongson3_cpucfg_data[0] = data;
+}
+
+static void patch_cpucfg_sel2(struct cpuinfo_mips *c)
+{
+	u64 ases = c->ases;
+	u64 options = c->options;
+	u32 data = c->loongson3_cpucfg_data[1];
+
+	if (ases & MIPS_ASE_LOONGSON_EXT)
+		data |= LOONGSON_CFG2_LEXT1;
+	if (ases & MIPS_ASE_LOONGSON_EXT2)
+		data |= LOONGSON_CFG2_LEXT2;
+	if (options & MIPS_CPU_LDPTE)
+		data |= LOONGSON_CFG2_LSPW;
+
+	if (ases & MIPS_ASE_VZ)
+		data |= LOONGSON_CFG2_LVZP;
+	else
+		data &= ~LOONGSON_CFG2_LVZREV;
+
+	c->loongson3_cpucfg_data[1] = data;
+}
+
+static void patch_cpucfg_sel3(struct cpuinfo_mips *c)
+{
+	u64 ases = c->ases;
+	u32 data = c->loongson3_cpucfg_data[2];
+
+	if (ases & MIPS_ASE_LOONGSON_CAM) {
+		data |= LOONGSON_CFG3_LCAMP;
+	} else {
+		data &= ~LOONGSON_CFG3_LCAMREV;
+		data &= ~LOONGSON_CFG3_LCAMNUM;
+		data &= ~LOONGSON_CFG3_LCAMKW;
+		data &= ~LOONGSON_CFG3_LCAMVW;
+	}
+
+	c->loongson3_cpucfg_data[2] = data;
+}
+
+void loongson3_cpucfg_finish_synthesis(struct cpuinfo_mips *c)
+{
+	/* CPUs with CPUCFG support don't need to synthesize anything. */
+	if (cpu_has_cfg())
+		return;
+
+	patch_cpucfg_sel1(c);
+	patch_cpucfg_sel2(c);
+	patch_cpucfg_sel3(c);
+}