* [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
@ 2020-08-06 12:26 Philippe Mathieu-Daudé
2020-08-06 18:01 ` Jiaxun Yang
0 siblings, 1 reply; 10+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-08-06 12:26 UTC (permalink / raw)
To: qemu-devel
Cc: Aleksandar Rikalo, Philippe Mathieu-Daudé,
Aleksandar Markovic, Huacai Chen, Aurelien Jarno
We only implement the Index[Store/Load]Tag from the 'cache' opcode.
Instead of ignoring the other cache operations, report them as
unimplemented.
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
---
target/mips/op_helper.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/target/mips/op_helper.c b/target/mips/op_helper.c
index 9552b280e0..7f87e57c8e 100644
--- a/target/mips/op_helper.c
+++ b/target/mips/op_helper.c
@@ -1583,6 +1583,8 @@ void helper_cache(CPUMIPSState *env, target_ulong addr, uint32_t op)
/* Index Load Tag */
memory_region_dispatch_read(env->itc_tag, index, &env->CP0_TagLo,
MO_64, MEMTXATTRS_UNSPECIFIED);
+ } else {
+ qemu_log_mask(LOG_UNIMP, "cache %u\n", op);
}
#endif
}
--
2.21.3
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-06 12:26 [PATCH-for-5.2] target/mips: Report unimplemented cache() operations Philippe Mathieu-Daudé
@ 2020-08-06 18:01 ` Jiaxun Yang
2020-08-06 20:11 ` Philippe Mathieu-Daudé
0 siblings, 1 reply; 10+ messages in thread
From: Jiaxun Yang @ 2020-08-06 18:01 UTC (permalink / raw)
To: Philippe Mathieu-Daudé, qemu-devel
Cc: Huacai Chen, Aleksandar Markovic, Aleksandar Rikalo, Aurelien Jarno
在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
> Instead of ignoring the other cache operations, report them as
> unimplemented.
Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
opcodes
in QEMU. Why do we log this?
Thanks.
- Jiaxun
>
> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> ---
> target/mips/op_helper.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/target/mips/op_helper.c b/target/mips/op_helper.c
> index 9552b280e0..7f87e57c8e 100644
> --- a/target/mips/op_helper.c
> +++ b/target/mips/op_helper.c
> @@ -1583,6 +1583,8 @@ void helper_cache(CPUMIPSState *env, target_ulong addr, uint32_t op)
> /* Index Load Tag */
> memory_region_dispatch_read(env->itc_tag, index, &env->CP0_TagLo,
> MO_64, MEMTXATTRS_UNSPECIFIED);
> + } else {
> + qemu_log_mask(LOG_UNIMP, "cache %u\n", op);
> }
> #endif
> }
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-06 18:01 ` Jiaxun Yang
@ 2020-08-06 20:11 ` Philippe Mathieu-Daudé
2020-08-06 20:51 ` Peter Maydell
0 siblings, 1 reply; 10+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-08-06 20:11 UTC (permalink / raw)
To: Jiaxun Yang, qemu-devel
Cc: Huacai Chen, Aleksandar Markovic, Aleksandar Rikalo, Aurelien Jarno
On 8/6/20 8:01 PM, Jiaxun Yang wrote:
> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>> Instead of ignoring the other cache operations, report them as
>> unimplemented.
>
> Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
> opcodes
> in QEMU. Why do we log this?
I'm noticed this code is run on Linux 3.3.8 (4KEc):
8880: 3082000f andi v0,a0,0xf
8884: 10800008 beqz a0,88a8
8888: 00a21021 addu v0,a1,v0
888c: 08002227 j 889c
8890: 00001821 move v1,zero
8894: bcf90000 cache 0x19,0(a3)
8898: 24630010 addiu v1,v1,16
889c: 0062302b sltu a2,v1,v0
88a0: 14c0fffc bnez a2,8894
88a4: 00833821 addu a3,a0,v1
88a8: 03e00008 jr ra
88ac: 00000000 nop
Why silently ignore the opcode is not implemented instead of logging it?
>
> Thanks.
>
> - Jiaxun
>
>>
>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> ---
>> target/mips/op_helper.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/target/mips/op_helper.c b/target/mips/op_helper.c
>> index 9552b280e0..7f87e57c8e 100644
>> --- a/target/mips/op_helper.c
>> +++ b/target/mips/op_helper.c
>> @@ -1583,6 +1583,8 @@ void helper_cache(CPUMIPSState *env,
>> target_ulong addr, uint32_t op)
>> /* Index Load Tag */
>> memory_region_dispatch_read(env->itc_tag, index,
>> &env->CP0_TagLo,
>> MO_64, MEMTXATTRS_UNSPECIFIED);
>> + } else {
>> + qemu_log_mask(LOG_UNIMP, "cache %u\n", op);
>> }
>> #endif
>> }
>
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-06 20:11 ` Philippe Mathieu-Daudé
@ 2020-08-06 20:51 ` Peter Maydell
2020-08-06 21:37 ` Philippe Mathieu-Daudé
2020-08-07 7:57 ` Jiaxun Yang
0 siblings, 2 replies; 10+ messages in thread
From: Peter Maydell @ 2020-08-06 20:51 UTC (permalink / raw)
To: Philippe Mathieu-Daudé
Cc: Aleksandar Rikalo, QEMU Developers, Aleksandar Markovic,
Huacai Chen, Aurelien Jarno
On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>
> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
> > 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
> >> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
> >> Instead of ignoring the other cache operations, report them as
> >> unimplemented.
> >
> > Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
> > opcodes
> > in QEMU. Why do we log this?
>
> I'm noticed this code is run on Linux 3.3.8 (4KEc):
>
> 8880: 3082000f andi v0,a0,0xf
> 8884: 10800008 beqz a0,88a8
> 8888: 00a21021 addu v0,a1,v0
> 888c: 08002227 j 889c
> 8890: 00001821 move v1,zero
> 8894: bcf90000 cache 0x19,0(a3)
> 8898: 24630010 addiu v1,v1,16
> 889c: 0062302b sltu a2,v1,v0
> 88a0: 14c0fffc bnez a2,8894
> 88a4: 00833821 addu a3,a0,v1
> 88a8: 03e00008 jr ra
> 88ac: 00000000 nop
>
> Why silently ignore the opcode is not implemented instead of logging it?
I think the question is whether the opcode is supposed to have
some behaviour which we're not implementing, or whether "no-op"
is the correct behaviour for it (which it usually is for
cache invalidate type operations; compare the way the Arm
cache ops like IC_IALLU are just ARM_CP_NOP ops).
thanks
-- PMM
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-06 20:51 ` Peter Maydell
@ 2020-08-06 21:37 ` Philippe Mathieu-Daudé
2020-08-10 17:21 ` Philippe Mathieu-Daudé
2020-08-07 7:57 ` Jiaxun Yang
1 sibling, 1 reply; 10+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-08-06 21:37 UTC (permalink / raw)
To: Peter Maydell
Cc: Huacai Chen, Aleksandar Rikalo, QEMU Developers, Aurelien Jarno,
Aleksandar Markovic
On 8/6/20 10:51 PM, Peter Maydell wrote:
> On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>>
>> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
>>> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>>>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>>>> Instead of ignoring the other cache operations, report them as
>>>> unimplemented.
>>>
>>> Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
>>> opcodes
>>> in QEMU. Why do we log this?
>>
>> I'm noticed this code is run on Linux 3.3.8 (4KEc):
>>
>> 8880: 3082000f andi v0,a0,0xf
>> 8884: 10800008 beqz a0,88a8
>> 8888: 00a21021 addu v0,a1,v0
>> 888c: 08002227 j 889c
>> 8890: 00001821 move v1,zero
>> 8894: bcf90000 cache 0x19,0(a3)
>> 8898: 24630010 addiu v1,v1,16
>> 889c: 0062302b sltu a2,v1,v0
>> 88a0: 14c0fffc bnez a2,8894
>> 88a4: 00833821 addu a3,a0,v1
>> 88a8: 03e00008 jr ra
>> 88ac: 00000000 nop
>>
>> Why silently ignore the opcode is not implemented instead of logging it?
>
> I think the question is whether the opcode is supposed to have
> some behaviour which we're not implementing, or whether "no-op"
> is the correct behaviour for it (which it usually is for
> cache invalidate type operations; compare the way the Arm
> cache ops like IC_IALLU are just ARM_CP_NOP ops).
OK now I understand better, thanks.
I haven't found useful information about this 0x19=25 opcode value.
On a r10k core it is listed as 'Hit Writeback Invalidate (D)' but here
this is a 4kEc. The address used is a SRAM shared with a embedded DSP
on the same SoC. From a RevEng PoV it is helpful to see there is a such
cache access, as I can separate better the peripheral involved.
I'm happy using a trace event instead.
Jiaxun, can you list me the list of opcodes QEMU can safely ignore from
the TCG emulation PoV? That way we can comment them in the code such:
switch (op) {
case 9:
/* Index Store Tag */
...
break;
case 5:
/* Index Load Tag */
...
break;
case X:
case Y:
case Z:
/* No-Op for QEMU */
...
break;
default:
qemu_log_mask(LOG_UNIMP, "cache %u\n", op);
}
Thanks,
Phil.
>
> thanks
> -- PMM
>
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-06 20:51 ` Peter Maydell
2020-08-06 21:37 ` Philippe Mathieu-Daudé
@ 2020-08-07 7:57 ` Jiaxun Yang
1 sibling, 0 replies; 10+ messages in thread
From: Jiaxun Yang @ 2020-08-07 7:57 UTC (permalink / raw)
To: Peter Maydell, Philippe Mathieu-Daudé
Cc: Huacai Chen, Aleksandar Markovic, Aleksandar Rikalo,
QEMU Developers, Aurelien Jarno
在 2020/8/7 上午4:51, Peter Maydell 写道:
> On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
>>> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>>>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>>>> Instead of ignoring the other cache operations, report them as
>>>> unimplemented.
>>> Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
>>> opcodes
>>> in QEMU. Why do we log this?
>> I'm noticed this code is run on Linux 3.3.8 (4KEc):
>>
>> 8880: 3082000f andi v0,a0,0xf
>> 8884: 10800008 beqz a0,88a8
>> 8888: 00a21021 addu v0,a1,v0
>> 888c: 08002227 j 889c
>> 8890: 00001821 move v1,zero
>> 8894: bcf90000 cache 0x19,0(a3)
>> 8898: 24630010 addiu v1,v1,16
>> 889c: 0062302b sltu a2,v1,v0
>> 88a0: 14c0fffc bnez a2,8894
>> 88a4: 00833821 addu a3,a0,v1
>> 88a8: 03e00008 jr ra
>> 88ac: 00000000 nop
>>
>> Why silently ignore the opcode is not implemented instead of logging it?
> I think the question is whether the opcode is supposed to have
> some behaviour which we're not implementing, or whether "no-op"
> is the correct behaviour for it (which it usually is for
> cache invalidate type operations; compare the way the Arm
> cache ops like IC_IALLU are just ARM_CP_NOP ops).
Probably we should skip Inv & WB opcode and log other undefined ops?
Otherwise the log may be flushed by Cache ops.
Thanks.
- Jiaxun
>
> thanks
> -- PMM
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-06 21:37 ` Philippe Mathieu-Daudé
@ 2020-08-10 17:21 ` Philippe Mathieu-Daudé
2020-08-13 10:58 ` Jiaxun Yang
2020-08-13 15:23 ` Jiaxun Yang
0 siblings, 2 replies; 10+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-08-10 17:21 UTC (permalink / raw)
To: Jiaxun Yang
Cc: Peter Maydell, Aleksandar Rikalo, QEMU Developers,
Aleksandar Markovic, Huacai Chen, Aurelien Jarno
On 8/6/20 11:37 PM, Philippe Mathieu-Daudé wrote:
> On 8/6/20 10:51 PM, Peter Maydell wrote:
>> On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>>>
>>> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
>>>> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>>>>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>>>>> Instead of ignoring the other cache operations, report them as
>>>>> unimplemented.
>>>>
>>>> Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
>>>> opcodes
>>>> in QEMU. Why do we log this?
>>>
>>> I'm noticed this code is run on Linux 3.3.8 (4KEc):
>>>
>>> 8880: 3082000f andi v0,a0,0xf
>>> 8884: 10800008 beqz a0,88a8
>>> 8888: 00a21021 addu v0,a1,v0
>>> 888c: 08002227 j 889c
>>> 8890: 00001821 move v1,zero
>>> 8894: bcf90000 cache 0x19,0(a3)
>>> 8898: 24630010 addiu v1,v1,16
>>> 889c: 0062302b sltu a2,v1,v0
>>> 88a0: 14c0fffc bnez a2,8894
>>> 88a4: 00833821 addu a3,a0,v1
>>> 88a8: 03e00008 jr ra
>>> 88ac: 00000000 nop
>>>
>>> Why silently ignore the opcode is not implemented instead of logging it?
>>
>> I think the question is whether the opcode is supposed to have
>> some behaviour which we're not implementing, or whether "no-op"
>> is the correct behaviour for it (which it usually is for
>> cache invalidate type operations; compare the way the Arm
>> cache ops like IC_IALLU are just ARM_CP_NOP ops).
>
> OK now I understand better, thanks.
>
> I haven't found useful information about this 0x19=25 opcode value.
Just to close this thread, some findings from last WE:
- I couldn't find where Linux 3.3.8 use that op
- I eventually figured out it comes from a kernel module called 'tiatm'.
- This kmod is released by OpenWRT in packages named kmod-sangam-atm-annex
- Googling for strings from the object, this file has been added in [1]
based on the file included in [2]
- Someone imported these files in a git repo and published
- There is a commented reference [4] as:
#define DataCacheHitInvalidate(a) {__asm__(" cache 17, (%0)"
: : "r" (a));}
#define DataCacheHitWriteback(a) {__asm__(" cache 25, (%0)"
: : "r" (a));}
- Also referenced (not commented) in [5] "Linux atm module implementation".
For my use I'm happy using a trace event:
-- >8 --
diff --git a/target/mips/op_helper.c b/target/mips/op_helper.c
index 7f87e57c8e..71b28ede2d 100644
--- a/target/mips/op_helper.c
+++ b/target/mips/op_helper.c
@@ -30,2 +30,3 @@
#include "sysemu/kvm.h"
+#include "trace.h"
@@ -1577,2 +1578,4 @@ void helper_cache(CPUMIPSState *env, target_ulong
addr, uint32_t op)
target_ulong index = addr & 0x1fffffff;
+
+ trace_cache_op(op, addr);
if (op == 9) {
diff --git a/target/mips/trace-events b/target/mips/trace-events
index ba87fe6062..8a60f23bbd 100644
--- a/target/mips/trace-events
+++ b/target/mips/trace-events
@@ -2,2 +2,5 @@
+# op_helper.c
+cache_op(uint32_t op, uint64_t addr) "cache op:%u paddr:0x%" PRIx64
+
# translate.c
---
[1]
https://git.openwrt.org/?p=openwrt/svn-archive/archive.git;a=commit;h=5a8a8f35c5a356f7167c3b3a3ca00f0780d86473
[2] https://dev.archive.openwrt.org/ticket/1411.html
[3] https://github.com/wolfhechel/ar7-atm
[4] https://github.com/wolfhechel/ar7-atm/blob/master/cpswhal_cpsar.h#L84
[5] https://github.com/wolfhechel/ar7-atm/blob/master/tn7atm.c#L479
>
> On a r10k core it is listed as 'Hit Writeback Invalidate (D)' but here
> this is a 4kEc. The address used is a SRAM shared with a embedded DSP
> on the same SoC. From a RevEng PoV it is helpful to see there is a such
> cache access, as I can separate better the peripheral involved.
> I'm happy using a trace event instead.
>
> Jiaxun, can you list me the list of opcodes QEMU can safely ignore from
> the TCG emulation PoV? That way we can comment them in the code such:
>
> switch (op) {
> case 9:
> /* Index Store Tag */
> ...
> break;
> case 5:
> /* Index Load Tag */
> ...
> break;
> case X:
> case Y:
> case Z:
> /* No-Op for QEMU */
> ...
> break;
> default:
> qemu_log_mask(LOG_UNIMP, "cache %u\n", op);
> }
>
> Thanks,
>
> Phil.
>
>>
>> thanks
>> -- PMM
>>
>
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-10 17:21 ` Philippe Mathieu-Daudé
@ 2020-08-13 10:58 ` Jiaxun Yang
2020-08-13 15:23 ` Jiaxun Yang
1 sibling, 0 replies; 10+ messages in thread
From: Jiaxun Yang @ 2020-08-13 10:58 UTC (permalink / raw)
To: Philippe Mathieu-Daudé
Cc: Peter Maydell, Aleksandar Rikalo, QEMU Developers,
Aleksandar Markovic, Huacai Chen, Aurelien Jarno
在 2020/8/11 上午1:21, Philippe Mathieu-Daudé 写道:
> On 8/6/20 11:37 PM, Philippe Mathieu-Daudé wrote:
>> On 8/6/20 10:51 PM, Peter Maydell wrote:
>>> On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>>>> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
>>>>> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>>>>>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>>>>>> Instead of ignoring the other cache operations, report them as
>>>>>> unimplemented.
[...]
>> On a r10k core it is listed as 'Hit Writeback Invalidate (D)' but here
>> this is a 4kEc. The address used is a SRAM shared with a embedded DSP
>> on the same SoC. From a RevEng PoV it is helpful to see there is a such
>> cache access, as I can separate better the peripheral involved.
>> I'm happy using a trace event instead.
>>
>> Jiaxun, can you list me the list of opcodes QEMU can safely ignore from
>> the TCG emulation PoV? That way we can comment them in the code such:
Hi Phil,
I believe we have nothing to do with all VA Hit based invalidate,
writeback, fetch and lock,
According to MD00086-2B-MIPS32BIS-AFP, the Code (Bit[20:18]) should be
0b100 (Hit Inavlidate), 0b101(Hit Fill), 0b110 (Hit FB), 0b111 (Fetch
and Lock).
I'm unsure about what applications expected from Index based ops, so
we'd better keep a
log for them. What I can say is Linux rarely use Index based ops.
Thanks.
- Jiaxun
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-10 17:21 ` Philippe Mathieu-Daudé
2020-08-13 10:58 ` Jiaxun Yang
@ 2020-08-13 15:23 ` Jiaxun Yang
2020-08-13 17:59 ` Philippe Mathieu-Daudé
1 sibling, 1 reply; 10+ messages in thread
From: Jiaxun Yang @ 2020-08-13 15:23 UTC (permalink / raw)
To: Philippe Mathieu-Daudé; +Cc: Aleksandar Markovic, QEMU Developers
在 2020/8/11 上午1:21, Philippe Mathieu-Daudé 写道:
> On 8/6/20 11:37 PM, Philippe Mathieu-Daudé wrote:
>> On 8/6/20 10:51 PM, Peter Maydell wrote:
>>> On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>>>> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
>>>>> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>>>>>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>>>>>> Instead of ignoring the other cache operations, report them as
>>>>>> unimplemented.
>>>>> Hmm, I don't think we have anything to do with Invalidate/Writeback etc.
>>>>> opcodes
>>>>> in QEMU. Why do we log this?
>>>> I'm noticed this code is run on Linux 3.3.8 (4KEc):
>>>>
>>>> 8880: 3082000f andi v0,a0,0xf
>>>> 8884: 10800008 beqz a0,88a8
>>>> 8888: 00a21021 addu v0,a1,v0
>>>> 888c: 08002227 j 889c
>>>> 8890: 00001821 move v1,zero
>>>> 8894: bcf90000 cache 0x19,0(a3)
>>>> 8898: 24630010 addiu v1,v1,16
>>>> 889c: 0062302b sltu a2,v1,v0
>>>> 88a0: 14c0fffc bnez a2,8894
>>>> 88a4: 00833821 addu a3,a0,v1
>>>> 88a8: 03e00008 jr ra
>>>> 88ac: 00000000 nop
>>>>
>>>> Why silently ignore the opcode is not implemented instead of logging it?
>>> I think the question is whether the opcode is supposed to have
>>> some behaviour which we're not implementing, or whether "no-op"
>>> is the correct behaviour for it (which it usually is for
>>> cache invalidate type operations; compare the way the Arm
>>> cache ops like IC_IALLU are just ARM_CP_NOP ops).
>> OK now I understand better, thanks.
>>
>> I haven't found useful information about this 0x19=25 opcode value.
> Just to close this thread, some findings from last WE:
>
> - I couldn't find where Linux 3.3.8 use that op
> - I eventually figured out it comes from a kernel module called 'tiatm'.
> - This kmod is released by OpenWRT in packages named kmod-sangam-atm-annex
> - Googling for strings from the object, this file has been added in [1]
> based on the file included in [2]
> - Someone imported these files in a git repo and published
> - There is a commented reference [4] as:
>
> #define DataCacheHitInvalidate(a) {__asm__(" cache 17, (%0)"
> : : "r" (a));}
> #define DataCacheHitWriteback(a) {__asm__(" cache 25, (%0)"
> : : "r" (a));}
>
> - Also referenced (not commented) in [5] "Linux atm module implementation".
>
> For my use I'm happy using a trace event:
>
> -- >8 --
FYI, I took a look at MIPS M14K (interAptiv) RTL from MIPSOpen project,
HitInvalidate and HitWriteback cache Ops to L2/L3 Cache will also act
on L1 Cache. So probably that's kinda... Undocumented trick...
Thanks.
- Jiaxun
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH-for-5.2] target/mips: Report unimplemented cache() operations
2020-08-13 15:23 ` Jiaxun Yang
@ 2020-08-13 17:59 ` Philippe Mathieu-Daudé
0 siblings, 0 replies; 10+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-08-13 17:59 UTC (permalink / raw)
To: Jiaxun Yang; +Cc: Aleksandar Markovic, QEMU Developers
On 8/13/20 5:23 PM, Jiaxun Yang wrote:
> 在 2020/8/11 上午1:21, Philippe Mathieu-Daudé 写道:
>> On 8/6/20 11:37 PM, Philippe Mathieu-Daudé wrote:
>>> On 8/6/20 10:51 PM, Peter Maydell wrote:
>>>> On Thu, 6 Aug 2020 at 21:31, Philippe Mathieu-Daudé
>>>> <f4bug@amsat.org> wrote:
>>>>> On 8/6/20 8:01 PM, Jiaxun Yang wrote:
>>>>>> 在 2020/8/6 下午8:26, Philippe Mathieu-Daudé 写道:
>>>>>>> We only implement the Index[Store/Load]Tag from the 'cache' opcode.
>>>>>>> Instead of ignoring the other cache operations, report them as
>>>>>>> unimplemented.
>>>>>> Hmm, I don't think we have anything to do with
>>>>>> Invalidate/Writeback etc.
>>>>>> opcodes
>>>>>> in QEMU. Why do we log this?
>>>>> I'm noticed this code is run on Linux 3.3.8 (4KEc):
>>>>>
>>>>> 8880: 3082000f andi v0,a0,0xf
>>>>> 8884: 10800008 beqz a0,88a8
>>>>> 8888: 00a21021 addu v0,a1,v0
>>>>> 888c: 08002227 j 889c
>>>>> 8890: 00001821 move v1,zero
>>>>> 8894: bcf90000 cache 0x19,0(a3)
>>>>> 8898: 24630010 addiu v1,v1,16
>>>>> 889c: 0062302b sltu a2,v1,v0
>>>>> 88a0: 14c0fffc bnez a2,8894
>>>>> 88a4: 00833821 addu a3,a0,v1
>>>>> 88a8: 03e00008 jr ra
>>>>> 88ac: 00000000 nop
>>>>>
>>>>> Why silently ignore the opcode is not implemented instead of
>>>>> logging it?
>>>> I think the question is whether the opcode is supposed to have
>>>> some behaviour which we're not implementing, or whether "no-op"
>>>> is the correct behaviour for it (which it usually is for
>>>> cache invalidate type operations; compare the way the Arm
>>>> cache ops like IC_IALLU are just ARM_CP_NOP ops).
>>> OK now I understand better, thanks.
>>>
>>> I haven't found useful information about this 0x19=25 opcode value.
>> Just to close this thread, some findings from last WE:
>>
>> - I couldn't find where Linux 3.3.8 use that op
>> - I eventually figured out it comes from a kernel module called 'tiatm'.
>> - This kmod is released by OpenWRT in packages named
>> kmod-sangam-atm-annex
>> - Googling for strings from the object, this file has been added in [1]
>> based on the file included in [2]
>> - Someone imported these files in a git repo and published
>> - There is a commented reference [4] as:
>>
>> #define DataCacheHitInvalidate(a) {__asm__(" cache 17, (%0)"
>> : : "r" (a));}
>> #define DataCacheHitWriteback(a) {__asm__(" cache 25, (%0)"
>> : : "r" (a));}
>>
>> - Also referenced (not commented) in [5] "Linux atm module
>> implementation".
>>
>> For my use I'm happy using a trace event:
>>
>> -- >8 --
>
> FYI, I took a look at MIPS M14K (interAptiv) RTL from MIPSOpen project,
> HitInvalidate and HitWriteback cache Ops to L2/L3 Cache will also act
> on L1 Cache. So probably that's kinda... Undocumented trick...
Uh... Thanks for having a look!
>
> Thanks.
>
> - Jiaxun
>
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2020-08-13 17:59 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-08-06 12:26 [PATCH-for-5.2] target/mips: Report unimplemented cache() operations Philippe Mathieu-Daudé
2020-08-06 18:01 ` Jiaxun Yang
2020-08-06 20:11 ` Philippe Mathieu-Daudé
2020-08-06 20:51 ` Peter Maydell
2020-08-06 21:37 ` Philippe Mathieu-Daudé
2020-08-10 17:21 ` Philippe Mathieu-Daudé
2020-08-13 10:58 ` Jiaxun Yang
2020-08-13 15:23 ` Jiaxun Yang
2020-08-13 17:59 ` Philippe Mathieu-Daudé
2020-08-07 7:57 ` Jiaxun Yang
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.