linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
@ 2022-11-08 12:58 Harshit Mogalapalli
  2022-11-09  7:05 ` Baolu Lu
  0 siblings, 1 reply; 12+ messages in thread
From: Harshit Mogalapalli @ 2022-11-08 12:58 UTC (permalink / raw)
  Cc: harshit.m.mogalapalli, konrad.wilk, harshit.m.mogalapalli,
	David Woodhouse, Lu Baolu, Joerg Roedel, Will Deacon,
	Robin Murphy, iommu, linux-kernel

It is likely that modern intel motherboard will not ship with a
floppy connection anymore, so let us disable it by default, as it
gets turned on when we do a make defconfig.

Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
---
 drivers/iommu/intel/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
index b7dff5092fd2..c783ae85ca9b 100644
--- a/drivers/iommu/intel/Kconfig
+++ b/drivers/iommu/intel/Kconfig
@@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
 	  option is removed in the 2.6.32 kernel.
 
 config INTEL_IOMMU_FLOPPY_WA
-	def_bool y
+	def_bool n
 	depends on X86
 	help
 	  Floppy disk drivers are known to bypass DMA API calls
-- 
2.38.1


^ permalink raw reply related	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-08 12:58 [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n Harshit Mogalapalli
@ 2022-11-09  7:05 ` Baolu Lu
  2022-11-09 12:16   ` Harshit Mogalapalli
  0 siblings, 1 reply; 12+ messages in thread
From: Baolu Lu @ 2022-11-09  7:05 UTC (permalink / raw)
  To: Harshit Mogalapalli
  Cc: baolu.lu, konrad.wilk, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, Robin Murphy, iommu, linux-kernel

On 2022/11/8 20:58, Harshit Mogalapalli wrote:
> It is likely that modern intel motherboard will not ship with a
> floppy connection anymore, so let us disable it by default, as it
> gets turned on when we do a make defconfig.
> 
> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
> ---
>   drivers/iommu/intel/Kconfig | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> index b7dff5092fd2..c783ae85ca9b 100644
> --- a/drivers/iommu/intel/Kconfig
> +++ b/drivers/iommu/intel/Kconfig
> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>   	  option is removed in the 2.6.32 kernel.
>   
>   config INTEL_IOMMU_FLOPPY_WA
> -	def_bool y
> +	def_bool n
>   	depends on X86
>   	help
>   	  Floppy disk drivers are known to bypass DMA API calls

Nobody selects or depends on this. How about removing this bool? Only
less than 10 lines of code are impacted and are not in any performance
path.

diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
index b7dff5092fd2..5e077d1c5f5d 100644
--- a/drivers/iommu/intel/Kconfig
+++ b/drivers/iommu/intel/Kconfig
@@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
  	  to use physical addresses for DMA, at least until this
  	  option is removed in the 2.6.32 kernel.

-config INTEL_IOMMU_FLOPPY_WA
-	def_bool y
-	depends on X86
-	help
-	  Floppy disk drivers are known to bypass DMA API calls
-	  thereby failing to work when IOMMU is enabled. This
-	  workaround will setup a 1:1 mapping for the first
-	  16MiB to make floppy (an ISA device) work.
-
  config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
  	bool "Enable Intel IOMMU scalable mode by default"
  	default y
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 48cdcd0a5cf3..22801850f339 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -4567,7 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
device *device,
  	}
  	rcu_read_unlock();

-#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
  	if (dev_is_pci(device)) {
  		struct pci_dev *pdev = to_pci_dev(device);

@@ -4579,7 +4578,6 @@ static void intel_iommu_get_resv_regions(struct 
device *device,
  				list_add_tail(&reg->list, head);
  		}
  	}
-#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */

  	reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
  				      IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,

Best regards,
baolu

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-09  7:05 ` Baolu Lu
@ 2022-11-09 12:16   ` Harshit Mogalapalli
  2022-11-09 13:16     ` Baolu Lu
  2022-11-09 13:26     ` Robin Murphy
  0 siblings, 2 replies; 12+ messages in thread
From: Harshit Mogalapalli @ 2022-11-09 12:16 UTC (permalink / raw)
  To: Baolu Lu
  Cc: konrad.wilk, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, Robin Murphy, iommu, linux-kernel,
	Darren Kenny, Harshit Mogalapalli



On 09/11/22 12:35 pm, Baolu Lu wrote:
> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>> It is likely that modern intel motherboard will not ship with a
>> floppy connection anymore, so let us disable it by default, as it
>> gets turned on when we do a make defconfig.
>>
>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
>> ---
>>   drivers/iommu/intel/Kconfig | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>> index b7dff5092fd2..c783ae85ca9b 100644
>> --- a/drivers/iommu/intel/Kconfig
>> +++ b/drivers/iommu/intel/Kconfig
>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>         option is removed in the 2.6.32 kernel.
>>   config INTEL_IOMMU_FLOPPY_WA
>> -    def_bool y
>> +    def_bool n
>>       depends on X86
>>       help
>>         Floppy disk drivers are known to bypass DMA API calls
> 
> Nobody selects or depends on this. How about removing this bool? Only
> less than 10 lines of code are impacted and are not in any performance
> path.
> 
> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> index b7dff5092fd2..5e077d1c5f5d 100644
> --- a/drivers/iommu/intel/Kconfig
> +++ b/drivers/iommu/intel/Kconfig
> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>         to use physical addresses for DMA, at least until this
>         option is removed in the 2.6.32 kernel.
> 
> -config INTEL_IOMMU_FLOPPY_WA
> -    def_bool y
> -    depends on X86
> -    help
> -      Floppy disk drivers are known to bypass DMA API calls
> -      thereby failing to work when IOMMU is enabled. This
> -      workaround will setup a 1:1 mapping for the first
> -      16MiB to make floppy (an ISA device) work.
> -
>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>       bool "Enable Intel IOMMU scalable mode by default"
>       default y
> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> index 48cdcd0a5cf3..22801850f339 100644
> --- a/drivers/iommu/intel/iommu.c
> +++ b/drivers/iommu/intel/iommu.c
> @@ -4567,7 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
> device *device,
>       }
>       rcu_read_unlock();
> 
> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>       if (dev_is_pci(device)) {
>           struct pci_dev *pdev = to_pci_dev(device);
> 
> @@ -4579,7 +4578,6 @@ static void intel_iommu_get_resv_regions(struct 
> device *device,
>                   list_add_tail(&reg->list, head);
>           }
>       }
> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> 
>       reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>                         IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
> 

Hi Baolu,

I have a question:
Shouldn't we remove the code between ifdef-endif statements?

I mean something like this:

diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
index b7dff5092fd2..5e077d1c5f5d 100644
--- a/drivers/iommu/intel/Kconfig
+++ b/drivers/iommu/intel/Kconfig
@@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
           to use physical addresses for DMA, at least until this
           option is removed in the 2.6.32 kernel.

-config INTEL_IOMMU_FLOPPY_WA
-       def_bool y
-       depends on X86
-       help
-         Floppy disk drivers are known to bypass DMA API calls
-         thereby failing to work when IOMMU is enabled. This
-         workaround will setup a 1:1 mapping for the first
-         16MiB to make floppy (an ISA device) work.
-
  config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
         bool "Enable Intel IOMMU scalable mode by default"
         default y
diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
index 48cdcd0a5cf3..2c416ad3204e 100644
--- a/drivers/iommu/intel/iommu.c
+++ b/drivers/iommu/intel/iommu.c
@@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
device *device,
         }
         rcu_read_unlock();

-#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
-       if (dev_is_pci(device)) {
-               struct pci_dev *pdev = to_pci_dev(device);
-
-               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
-                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
-                                       IOMMU_RESV_DIRECT_RELAXABLE,
-                                       GFP_KERNEL);
-                       if (reg)
-                               list_add_tail(&reg->list, head);
-               }
-       }
-#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
-
         reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
                                       IOAPIC_RANGE_END - 
IOAPIC_RANGE_START + 1,
                                       0, IOMMU_RESV_MSI, GFP_KERNEL);

This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA 
direct mapping region via iommu_get_resv_regions")

If others don't have an objection I can send a patch with the above diff.

Note: I am unsure about the code internals.

Thanks,
Harshit


> Best regards,
> baolu

^ permalink raw reply related	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-09 12:16   ` Harshit Mogalapalli
@ 2022-11-09 13:16     ` Baolu Lu
  2022-11-09 20:17       ` Konrad Rzeszutek Wilk
  2022-11-09 13:26     ` Robin Murphy
  1 sibling, 1 reply; 12+ messages in thread
From: Baolu Lu @ 2022-11-09 13:16 UTC (permalink / raw)
  To: Harshit Mogalapalli
  Cc: baolu.lu, konrad.wilk, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, Robin Murphy, iommu, linux-kernel,
	Darren Kenny

On 2022/11/9 20:16, Harshit Mogalapalli wrote:
> 
> 
> On 09/11/22 12:35 pm, Baolu Lu wrote:
>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>> It is likely that modern intel motherboard will not ship with a
>>> floppy connection anymore, so let us disable it by default, as it
>>> gets turned on when we do a make defconfig.
>>>
>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
>>> ---
>>>   drivers/iommu/intel/Kconfig | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>> index b7dff5092fd2..c783ae85ca9b 100644
>>> --- a/drivers/iommu/intel/Kconfig
>>> +++ b/drivers/iommu/intel/Kconfig
>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>         option is removed in the 2.6.32 kernel.
>>>   config INTEL_IOMMU_FLOPPY_WA
>>> -    def_bool y
>>> +    def_bool n
>>>       depends on X86
>>>       help
>>>         Floppy disk drivers are known to bypass DMA API calls
>>
>> Nobody selects or depends on this. How about removing this bool? Only
>> less than 10 lines of code are impacted and are not in any performance
>> path.
>>
>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>> index b7dff5092fd2..5e077d1c5f5d 100644
>> --- a/drivers/iommu/intel/Kconfig
>> +++ b/drivers/iommu/intel/Kconfig
>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>         to use physical addresses for DMA, at least until this
>>         option is removed in the 2.6.32 kernel.
>>
>> -config INTEL_IOMMU_FLOPPY_WA
>> -    def_bool y
>> -    depends on X86
>> -    help
>> -      Floppy disk drivers are known to bypass DMA API calls
>> -      thereby failing to work when IOMMU is enabled. This
>> -      workaround will setup a 1:1 mapping for the first
>> -      16MiB to make floppy (an ISA device) work.
>> -
>>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>       bool "Enable Intel IOMMU scalable mode by default"
>>       default y
>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>> index 48cdcd0a5cf3..22801850f339 100644
>> --- a/drivers/iommu/intel/iommu.c
>> +++ b/drivers/iommu/intel/iommu.c
>> @@ -4567,7 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
>> device *device,
>>       }
>>       rcu_read_unlock();
>>
>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>       if (dev_is_pci(device)) {
>>           struct pci_dev *pdev = to_pci_dev(device);
>>
>> @@ -4579,7 +4578,6 @@ static void intel_iommu_get_resv_regions(struct 
>> device *device,
>>                   list_add_tail(&reg->list, head);
>>           }
>>       }
>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>
>>       reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>                         IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>
> 
> Hi Baolu,
> 
> I have a question:
> Shouldn't we remove the code between ifdef-endif statements?
> 
> I mean something like this:
> 
> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> index b7dff5092fd2..5e077d1c5f5d 100644
> --- a/drivers/iommu/intel/Kconfig
> +++ b/drivers/iommu/intel/Kconfig
> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>            to use physical addresses for DMA, at least until this
>            option is removed in the 2.6.32 kernel.
> 
> -config INTEL_IOMMU_FLOPPY_WA
> -       def_bool y
> -       depends on X86
> -       help
> -         Floppy disk drivers are known to bypass DMA API calls
> -         thereby failing to work when IOMMU is enabled. This
> -         workaround will setup a 1:1 mapping for the first
> -         16MiB to make floppy (an ISA device) work.
> -
>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>          bool "Enable Intel IOMMU scalable mode by default"
>          default y
> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> index 48cdcd0a5cf3..2c416ad3204e 100644
> --- a/drivers/iommu/intel/iommu.c
> +++ b/drivers/iommu/intel/iommu.c
> @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
> device *device,
>          }
>          rcu_read_unlock();
> 
> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
> -       if (dev_is_pci(device)) {
> -               struct pci_dev *pdev = to_pci_dev(device);
> -
> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
> -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
> -                                       GFP_KERNEL);
> -                       if (reg)
> -                               list_add_tail(&reg->list, head);
> -               }
> -       }
> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> -
>          reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>                                        IOAPIC_RANGE_END - 
> IOAPIC_RANGE_START + 1,
>                                        0, IOMMU_RESV_MSI, GFP_KERNEL);
> 
> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA 
> direct mapping region via iommu_get_resv_regions")

As long as floppy driver exists in the tree, we have to include above
code. Otherwise, floppy drivers don't work. At least we can easily find
drivers/block/floppy.c which is still maintained (check MAINTAINERS).:-)

Best regards,
baolu

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-09 12:16   ` Harshit Mogalapalli
  2022-11-09 13:16     ` Baolu Lu
@ 2022-11-09 13:26     ` Robin Murphy
  2022-11-09 13:55       ` Harshit Mogalapalli
  1 sibling, 1 reply; 12+ messages in thread
From: Robin Murphy @ 2022-11-09 13:26 UTC (permalink / raw)
  To: Harshit Mogalapalli, Baolu Lu
  Cc: konrad.wilk, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, iommu, linux-kernel, Darren Kenny

On 09/11/2022 12:16 pm, Harshit Mogalapalli wrote:
> 
> 
> On 09/11/22 12:35 pm, Baolu Lu wrote:
>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>> It is likely that modern intel motherboard will not ship with a
>>> floppy connection anymore, so let us disable it by default, as it
>>> gets turned on when we do a make defconfig.
>>>
>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
>>> ---
>>>   drivers/iommu/intel/Kconfig | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>> index b7dff5092fd2..c783ae85ca9b 100644
>>> --- a/drivers/iommu/intel/Kconfig
>>> +++ b/drivers/iommu/intel/Kconfig
>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>         option is removed in the 2.6.32 kernel.
>>>   config INTEL_IOMMU_FLOPPY_WA
>>> -    def_bool y
>>> +    def_bool n
>>>       depends on X86
>>>       help
>>>         Floppy disk drivers are known to bypass DMA API calls
>>
>> Nobody selects or depends on this. How about removing this bool? Only
>> less than 10 lines of code are impacted and are not in any performance
>> path.
>>
>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>> index b7dff5092fd2..5e077d1c5f5d 100644
>> --- a/drivers/iommu/intel/Kconfig
>> +++ b/drivers/iommu/intel/Kconfig
>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>         to use physical addresses for DMA, at least until this
>>         option is removed in the 2.6.32 kernel.
>>
>> -config INTEL_IOMMU_FLOPPY_WA
>> -    def_bool y
>> -    depends on X86
>> -    help
>> -      Floppy disk drivers are known to bypass DMA API calls
>> -      thereby failing to work when IOMMU is enabled. This
>> -      workaround will setup a 1:1 mapping for the first
>> -      16MiB to make floppy (an ISA device) work.
>> -
>>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>       bool "Enable Intel IOMMU scalable mode by default"
>>       default y
>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>> index 48cdcd0a5cf3..22801850f339 100644
>> --- a/drivers/iommu/intel/iommu.c
>> +++ b/drivers/iommu/intel/iommu.c
>> @@ -4567,7 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
>> device *device,
>>       }
>>       rcu_read_unlock();
>>
>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>       if (dev_is_pci(device)) {
>>           struct pci_dev *pdev = to_pci_dev(device);
>>
>> @@ -4579,7 +4578,6 @@ static void intel_iommu_get_resv_regions(struct 
>> device *device,
>>                   list_add_tail(&reg->list, head);
>>           }
>>       }
>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>
>>       reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>                         IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>
> 
> Hi Baolu,
> 
> I have a question:
> Shouldn't we remove the code between ifdef-endif statements?

No, machines with floppy drives may be uncommon, but almost certainly 
some are still in use and being updated with new kernels, so we should 
continue to support them.

Do you have any practical issues with the ISA reserved region being 
present, or is the only motivation here to save a few bytes of code when 
it won't be needed? In the latter case, making it "def_bool BLK_DEV_FD" 
might be sensible.

(For the original patch, note that "def_bool n" is always a bit 
redundant since plain "bool" implicitly defaults to "n" anyway)

Thanks,
Robin.

> I mean something like this:
> 
> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> index b7dff5092fd2..5e077d1c5f5d 100644
> --- a/drivers/iommu/intel/Kconfig
> +++ b/drivers/iommu/intel/Kconfig
> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>            to use physical addresses for DMA, at least until this
>            option is removed in the 2.6.32 kernel.
> 
> -config INTEL_IOMMU_FLOPPY_WA
> -       def_bool y
> -       depends on X86
> -       help
> -         Floppy disk drivers are known to bypass DMA API calls
> -         thereby failing to work when IOMMU is enabled. This
> -         workaround will setup a 1:1 mapping for the first
> -         16MiB to make floppy (an ISA device) work.
> -
>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>          bool "Enable Intel IOMMU scalable mode by default"
>          default y
> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> index 48cdcd0a5cf3..2c416ad3204e 100644
> --- a/drivers/iommu/intel/iommu.c
> +++ b/drivers/iommu/intel/iommu.c
> @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
> device *device,
>          }
>          rcu_read_unlock();
> 
> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
> -       if (dev_is_pci(device)) {
> -               struct pci_dev *pdev = to_pci_dev(device);
> -
> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
> -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
> -                                       GFP_KERNEL);
> -                       if (reg)
> -                               list_add_tail(&reg->list, head);
> -               }
> -       }
> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> -
>          reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>                                        IOAPIC_RANGE_END - 
> IOAPIC_RANGE_START + 1,
>                                        0, IOMMU_RESV_MSI, GFP_KERNEL);
> 
> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA 
> direct mapping region via iommu_get_resv_regions")
> 
> If others don't have an objection I can send a patch with the above diff.
> 
> Note: I am unsure about the code internals.
> 
> Thanks,
> Harshit
> 
> 
>> Best regards,
>> baolu

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-09 13:26     ` Robin Murphy
@ 2022-11-09 13:55       ` Harshit Mogalapalli
  0 siblings, 0 replies; 12+ messages in thread
From: Harshit Mogalapalli @ 2022-11-09 13:55 UTC (permalink / raw)
  To: Robin Murphy, Baolu Lu
  Cc: konrad.wilk, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, iommu, linux-kernel, Darren Kenny,
	Harshit Mogalapalli



On 09/11/22 6:56 pm, Robin Murphy wrote:
> On 09/11/2022 12:16 pm, Harshit Mogalapalli wrote:
>>
>>
>> On 09/11/22 12:35 pm, Baolu Lu wrote:
>>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>>> It is likely that modern intel motherboard will not ship with a
>>>> floppy connection anymore, so let us disable it by default, as it
>>>> gets turned on when we do a make defconfig.
>>>>
>>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>>> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
>>>> ---
>>>>   drivers/iommu/intel/Kconfig | 2 +-
>>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>>> index b7dff5092fd2..c783ae85ca9b 100644
>>>> --- a/drivers/iommu/intel/Kconfig
>>>> +++ b/drivers/iommu/intel/Kconfig
>>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>         option is removed in the 2.6.32 kernel.
>>>>   config INTEL_IOMMU_FLOPPY_WA
>>>> -    def_bool y
>>>> +    def_bool n
>>>>       depends on X86
>>>>       help
>>>>         Floppy disk drivers are known to bypass DMA API calls
>>>
>>> Nobody selects or depends on this. How about removing this bool? Only
>>> less than 10 lines of code are impacted and are not in any performance
>>> path.
>>>
>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>> --- a/drivers/iommu/intel/Kconfig
>>> +++ b/drivers/iommu/intel/Kconfig
>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>         to use physical addresses for DMA, at least until this
>>>         option is removed in the 2.6.32 kernel.
>>>
>>> -config INTEL_IOMMU_FLOPPY_WA
>>> -    def_bool y
>>> -    depends on X86
>>> -    help
>>> -      Floppy disk drivers are known to bypass DMA API calls
>>> -      thereby failing to work when IOMMU is enabled. This
>>> -      workaround will setup a 1:1 mapping for the first
>>> -      16MiB to make floppy (an ISA device) work.
>>> -
>>>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>       bool "Enable Intel IOMMU scalable mode by default"
>>>       default y
>>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>>> index 48cdcd0a5cf3..22801850f339 100644
>>> --- a/drivers/iommu/intel/iommu.c
>>> +++ b/drivers/iommu/intel/iommu.c
>>> @@ -4567,7 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
>>> device *device,
>>>       }
>>>       rcu_read_unlock();
>>>
>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>       if (dev_is_pci(device)) {
>>>           struct pci_dev *pdev = to_pci_dev(device);
>>>
>>> @@ -4579,7 +4578,6 @@ static void intel_iommu_get_resv_regions(struct 
>>> device *device,
>>>                   list_add_tail(&reg->list, head);
>>>           }
>>>       }
>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>
>>>       reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>                         IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>>
>>
>> Hi Baolu,
>>
>> I have a question:
>> Shouldn't we remove the code between ifdef-endif statements?
> 
Hi Baolu and Robin,


> No, machines with floppy drives may be uncommon, but almost certainly 
> some are still in use and being updated with new kernels, so we should 
> continue to support them.
> 
Thanks a lot for explaining.

I misunderstood the code, I thought when we are removing ifdef-endif 
statement, the code inside(within ifdef-endif) becomes redundant, thanks 
for correcting me.

> Do you have any practical issues with the ISA reserved region being 
> present, or is the only motivation here to save a few bytes of code when 
> it won't be needed? In the latter case, making it "def_bool BLK_DEV_FD" 
> might be sensible.
>

No practical issues seen.

> (For the original patch, note that "def_bool n" is always a bit 
> redundant since plain "bool" implicitly defaults to "n" anyway)
> 
Okay.
How about Baolu's version of the patch, where we are removing the CONFIG 
item from Kconfig and ifdef statement from the code in iommu.c?

Thanks,
Harshit


> Thanks,
> Robin.
> 
>> I mean something like this:
>>
>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>> index b7dff5092fd2..5e077d1c5f5d 100644
>> --- a/drivers/iommu/intel/Kconfig
>> +++ b/drivers/iommu/intel/Kconfig
>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>            to use physical addresses for DMA, at least until this
>>            option is removed in the 2.6.32 kernel.
>>
>> -config INTEL_IOMMU_FLOPPY_WA
>> -       def_bool y
>> -       depends on X86
>> -       help
>> -         Floppy disk drivers are known to bypass DMA API calls
>> -         thereby failing to work when IOMMU is enabled. This
>> -         workaround will setup a 1:1 mapping for the first
>> -         16MiB to make floppy (an ISA device) work.
>> -
>>   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>          bool "Enable Intel IOMMU scalable mode by default"
>>          default y
>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>> index 48cdcd0a5cf3..2c416ad3204e 100644
>> --- a/drivers/iommu/intel/iommu.c
>> +++ b/drivers/iommu/intel/iommu.c
>> @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct 
>> device *device,
>>          }
>>          rcu_read_unlock();
>>
>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>> -       if (dev_is_pci(device)) {
>> -               struct pci_dev *pdev = to_pci_dev(device);
>> -
>> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
>> -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
>> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
>> -                                       GFP_KERNEL);
>> -                       if (reg)
>> -                               list_add_tail(&reg->list, head);
>> -               }
>> -       }
>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>> -
>>          reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>                                        IOAPIC_RANGE_END - 
>> IOAPIC_RANGE_START + 1,
>>                                        0, IOMMU_RESV_MSI, GFP_KERNEL);
>>
>> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose 
>> ISA direct mapping region via iommu_get_resv_regions")
>>
>> If others don't have an objection I can send a patch with the above diff.
>>
>> Note: I am unsure about the code internals.
>>
>> Thanks,
>> Harshit
>>
>>
>>> Best regards,
>>> baolu

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-09 13:16     ` Baolu Lu
@ 2022-11-09 20:17       ` Konrad Rzeszutek Wilk
  2022-11-10  6:39         ` Baolu Lu
  0 siblings, 1 reply; 12+ messages in thread
From: Konrad Rzeszutek Wilk @ 2022-11-09 20:17 UTC (permalink / raw)
  To: Baolu Lu
  Cc: Harshit Mogalapalli, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, Robin Murphy, iommu, linux-kernel,
	Darren Kenny

On Wed, Nov 09, 2022 at 09:16:53PM +0800, Baolu Lu wrote:
> On 2022/11/9 20:16, Harshit Mogalapalli wrote:
> > 
> > 
> > On 09/11/22 12:35 pm, Baolu Lu wrote:
> > > On 2022/11/8 20:58, Harshit Mogalapalli wrote:
> > > > It is likely that modern intel motherboard will not ship with a
> > > > floppy connection anymore, so let us disable it by default, as it
> > > > gets turned on when we do a make defconfig.
> > > > 
> > > > Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> > > > Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
> > > > ---
> > > >   drivers/iommu/intel/Kconfig | 2 +-
> > > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > > 
> > > > diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> > > > index b7dff5092fd2..c783ae85ca9b 100644
> > > > --- a/drivers/iommu/intel/Kconfig
> > > > +++ b/drivers/iommu/intel/Kconfig
> > > > @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
> > > >         option is removed in the 2.6.32 kernel.
> > > >   config INTEL_IOMMU_FLOPPY_WA
> > > > -    def_bool y
> > > > +    def_bool n
> > > >       depends on X86
> > > >       help
> > > >         Floppy disk drivers are known to bypass DMA API calls
> > > 
> > > Nobody selects or depends on this. How about removing this bool? Only
> > > less than 10 lines of code are impacted and are not in any performance
> > > path.
> > > 
> > > diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> > > index b7dff5092fd2..5e077d1c5f5d 100644
> > > --- a/drivers/iommu/intel/Kconfig
> > > +++ b/drivers/iommu/intel/Kconfig
> > > @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
> > >         to use physical addresses for DMA, at least until this
> > >         option is removed in the 2.6.32 kernel.
> > > 
> > > -config INTEL_IOMMU_FLOPPY_WA
> > > -    def_bool y
> > > -    depends on X86
> > > -    help
> > > -      Floppy disk drivers are known to bypass DMA API calls
> > > -      thereby failing to work when IOMMU is enabled. This
> > > -      workaround will setup a 1:1 mapping for the first
> > > -      16MiB to make floppy (an ISA device) work.
> > > -
> > >   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
> > >       bool "Enable Intel IOMMU scalable mode by default"
> > >       default y
> > > diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> > > index 48cdcd0a5cf3..22801850f339 100644
> > > --- a/drivers/iommu/intel/iommu.c
> > > +++ b/drivers/iommu/intel/iommu.c
> > > @@ -4567,7 +4567,6 @@ static void
> > > intel_iommu_get_resv_regions(struct device *device,
> > >       }
> > >       rcu_read_unlock();
> > > 
> > > -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
> > >       if (dev_is_pci(device)) {
> > >           struct pci_dev *pdev = to_pci_dev(device);
> > > 
> > > @@ -4579,7 +4578,6 @@ static void
> > > intel_iommu_get_resv_regions(struct device *device,
> > >                   list_add_tail(&reg->list, head);
> > >           }
> > >       }
> > > -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> > > 
> > >       reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
> > >                         IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
> > > 
> > 
> > Hi Baolu,
> > 
> > I have a question:
> > Shouldn't we remove the code between ifdef-endif statements?
> > 
> > I mean something like this:
> > 
> > diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> > index b7dff5092fd2..5e077d1c5f5d 100644
> > --- a/drivers/iommu/intel/Kconfig
> > +++ b/drivers/iommu/intel/Kconfig
> > @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
> >            to use physical addresses for DMA, at least until this
> >            option is removed in the 2.6.32 kernel.
> > 
> > -config INTEL_IOMMU_FLOPPY_WA
> > -       def_bool y
> > -       depends on X86
> > -       help
> > -         Floppy disk drivers are known to bypass DMA API calls
> > -         thereby failing to work when IOMMU is enabled. This
> > -         workaround will setup a 1:1 mapping for the first
> > -         16MiB to make floppy (an ISA device) work.
> > -
> >   config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
> >          bool "Enable Intel IOMMU scalable mode by default"
> >          default y
> > diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> > index 48cdcd0a5cf3..2c416ad3204e 100644
> > --- a/drivers/iommu/intel/iommu.c
> > +++ b/drivers/iommu/intel/iommu.c
> > @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct
> > device *device,
> >          }
> >          rcu_read_unlock();
> > 
> > -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
> > -       if (dev_is_pci(device)) {
> > -               struct pci_dev *pdev = to_pci_dev(device);
> > -
> > -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
> > -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
> > -                                       IOMMU_RESV_DIRECT_RELAXABLE,
> > -                                       GFP_KERNEL);
> > -                       if (reg)
> > -                               list_add_tail(&reg->list, head);
> > -               }
> > -       }
> > -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> > -
> >          reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
> >                                        IOAPIC_RANGE_END -
> > IOAPIC_RANGE_START + 1,
> >                                        0, IOMMU_RESV_MSI, GFP_KERNEL);
> > 
> > This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA
> > direct mapping region via iommu_get_resv_regions")
> 
> As long as floppy driver exists in the tree, we have to include above
> code. Otherwise, floppy drivers don't work. At least we can easily find
> drivers/block/floppy.c which is still maintained (check MAINTAINERS).:-)

But this requires a machine with Intel IOMMU and ISA:

> > > -      16MiB to make floppy (an ISA device) work.

ISA device? I don't believe there are any Intel machines with an IOMMU
and an ISA device?

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-09 20:17       ` Konrad Rzeszutek Wilk
@ 2022-11-10  6:39         ` Baolu Lu
  2022-11-10 21:00           ` Konrad Rzeszutek Wilk
  0 siblings, 1 reply; 12+ messages in thread
From: Baolu Lu @ 2022-11-10  6:39 UTC (permalink / raw)
  To: Konrad Rzeszutek Wilk
  Cc: baolu.lu, Harshit Mogalapalli, harshit.m.mogalapalli,
	David Woodhouse, Joerg Roedel, Will Deacon, Robin Murphy, iommu,
	linux-kernel, Darren Kenny

On 2022/11/10 4:17, Konrad Rzeszutek Wilk wrote:
> On Wed, Nov 09, 2022 at 09:16:53PM +0800, Baolu Lu wrote:
>> On 2022/11/9 20:16, Harshit Mogalapalli wrote:
>>>
>>>
>>> On 09/11/22 12:35 pm, Baolu Lu wrote:
>>>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>>>> It is likely that modern intel motherboard will not ship with a
>>>>> floppy connection anymore, so let us disable it by default, as it
>>>>> gets turned on when we do a make defconfig.
>>>>>
>>>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>>>> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
>>>>> ---
>>>>>    drivers/iommu/intel/Kconfig | 2 +-
>>>>>    1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>>>> index b7dff5092fd2..c783ae85ca9b 100644
>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>          option is removed in the 2.6.32 kernel.
>>>>>    config INTEL_IOMMU_FLOPPY_WA
>>>>> -    def_bool y
>>>>> +    def_bool n
>>>>>        depends on X86
>>>>>        help
>>>>>          Floppy disk drivers are known to bypass DMA API calls
>>>>
>>>> Nobody selects or depends on this. How about removing this bool? Only
>>>> less than 10 lines of code are impacted and are not in any performance
>>>> path.
>>>>
>>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>> --- a/drivers/iommu/intel/Kconfig
>>>> +++ b/drivers/iommu/intel/Kconfig
>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>          to use physical addresses for DMA, at least until this
>>>>          option is removed in the 2.6.32 kernel.
>>>>
>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>> -    def_bool y
>>>> -    depends on X86
>>>> -    help
>>>> -      Floppy disk drivers are known to bypass DMA API calls
>>>> -      thereby failing to work when IOMMU is enabled. This
>>>> -      workaround will setup a 1:1 mapping for the first
>>>> -      16MiB to make floppy (an ISA device) work.
>>>> -
>>>>    config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>        bool "Enable Intel IOMMU scalable mode by default"
>>>>        default y
>>>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>>>> index 48cdcd0a5cf3..22801850f339 100644
>>>> --- a/drivers/iommu/intel/iommu.c
>>>> +++ b/drivers/iommu/intel/iommu.c
>>>> @@ -4567,7 +4567,6 @@ static void
>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>        }
>>>>        rcu_read_unlock();
>>>>
>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>        if (dev_is_pci(device)) {
>>>>            struct pci_dev *pdev = to_pci_dev(device);
>>>>
>>>> @@ -4579,7 +4578,6 @@ static void
>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>                    list_add_tail(&reg->list, head);
>>>>            }
>>>>        }
>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>
>>>>        reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>                          IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>>>
>>>
>>> Hi Baolu,
>>>
>>> I have a question:
>>> Shouldn't we remove the code between ifdef-endif statements?
>>>
>>> I mean something like this:
>>>
>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>> --- a/drivers/iommu/intel/Kconfig
>>> +++ b/drivers/iommu/intel/Kconfig
>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>             to use physical addresses for DMA, at least until this
>>>             option is removed in the 2.6.32 kernel.
>>>
>>> -config INTEL_IOMMU_FLOPPY_WA
>>> -       def_bool y
>>> -       depends on X86
>>> -       help
>>> -         Floppy disk drivers are known to bypass DMA API calls
>>> -         thereby failing to work when IOMMU is enabled. This
>>> -         workaround will setup a 1:1 mapping for the first
>>> -         16MiB to make floppy (an ISA device) work.
>>> -
>>>    config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>           bool "Enable Intel IOMMU scalable mode by default"
>>>           default y
>>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>>> index 48cdcd0a5cf3..2c416ad3204e 100644
>>> --- a/drivers/iommu/intel/iommu.c
>>> +++ b/drivers/iommu/intel/iommu.c
>>> @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct
>>> device *device,
>>>           }
>>>           rcu_read_unlock();
>>>
>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>> -       if (dev_is_pci(device)) {
>>> -               struct pci_dev *pdev = to_pci_dev(device);
>>> -
>>> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
>>> -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
>>> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
>>> -                                       GFP_KERNEL);
>>> -                       if (reg)
>>> -                               list_add_tail(&reg->list, head);
>>> -               }
>>> -       }
>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>> -
>>>           reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>                                         IOAPIC_RANGE_END -
>>> IOAPIC_RANGE_START + 1,
>>>                                         0, IOMMU_RESV_MSI, GFP_KERNEL);
>>>
>>> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA
>>> direct mapping region via iommu_get_resv_regions")
>>
>> As long as floppy driver exists in the tree, we have to include above
>> code. Otherwise, floppy drivers don't work. At least we can easily find
>> drivers/block/floppy.c which is still maintained (check MAINTAINERS).:-)
> 
> But this requires a machine with Intel IOMMU and ISA:
> 
>>>> -      16MiB to make floppy (an ISA device) work.
> 
> ISA device? I don't believe there are any Intel machines with an IOMMU
> and an ISA device?

This workaround was introduced by commit 49a0429e53f2 ("Intel IOMMU:
Iommu floppy workaround") in 2007. I can't remember what happened 15
years ago, but I believe there must have been corresponding hardware
configurations at that time, and the Linux kernel has been maintained it
to now.

Best regards,
baolu

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-10  6:39         ` Baolu Lu
@ 2022-11-10 21:00           ` Konrad Rzeszutek Wilk
  2022-11-11  2:31             ` Baolu Lu
  0 siblings, 1 reply; 12+ messages in thread
From: Konrad Rzeszutek Wilk @ 2022-11-10 21:00 UTC (permalink / raw)
  To: Baolu Lu
  Cc: Harshit Mogalapalli, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, Robin Murphy, iommu, linux-kernel,
	Darren Kenny

On Thu, Nov 10, 2022 at 02:39:53PM +0800, Baolu Lu wrote:
> On 2022/11/10 4:17, Konrad Rzeszutek Wilk wrote:
> > On Wed, Nov 09, 2022 at 09:16:53PM +0800, Baolu Lu wrote:
> > > On 2022/11/9 20:16, Harshit Mogalapalli wrote:
> > > > 
> > > > 
> > > > On 09/11/22 12:35 pm, Baolu Lu wrote:
> > > > > On 2022/11/8 20:58, Harshit Mogalapalli wrote:
> > > > > > It is likely that modern intel motherboard will not ship with a
> > > > > > floppy connection anymore, so let us disable it by default, as it
> > > > > > gets turned on when we do a make defconfig.
> > > > > > 
> > > > > > Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> > > > > > Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
> > > > > > ---
> > > > > >    drivers/iommu/intel/Kconfig | 2 +-
> > > > > >    1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > > 
> > > > > > diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> > > > > > index b7dff5092fd2..c783ae85ca9b 100644
> > > > > > --- a/drivers/iommu/intel/Kconfig
> > > > > > +++ b/drivers/iommu/intel/Kconfig
> > > > > > @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
> > > > > >          option is removed in the 2.6.32 kernel.
> > > > > >    config INTEL_IOMMU_FLOPPY_WA
> > > > > > -    def_bool y
> > > > > > +    def_bool n
> > > > > >        depends on X86
> > > > > >        help
> > > > > >          Floppy disk drivers are known to bypass DMA API calls
> > > > > 
> > > > > Nobody selects or depends on this. How about removing this bool? Only
> > > > > less than 10 lines of code are impacted and are not in any performance
> > > > > path.
> > > > > 
> > > > > diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> > > > > index b7dff5092fd2..5e077d1c5f5d 100644
> > > > > --- a/drivers/iommu/intel/Kconfig
> > > > > +++ b/drivers/iommu/intel/Kconfig
> > > > > @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
> > > > >          to use physical addresses for DMA, at least until this
> > > > >          option is removed in the 2.6.32 kernel.
> > > > > 
> > > > > -config INTEL_IOMMU_FLOPPY_WA
> > > > > -    def_bool y
> > > > > -    depends on X86
> > > > > -    help
> > > > > -      Floppy disk drivers are known to bypass DMA API calls
> > > > > -      thereby failing to work when IOMMU is enabled. This
> > > > > -      workaround will setup a 1:1 mapping for the first
> > > > > -      16MiB to make floppy (an ISA device) work.
> > > > > -
> > > > >    config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
> > > > >        bool "Enable Intel IOMMU scalable mode by default"
> > > > >        default y
> > > > > diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> > > > > index 48cdcd0a5cf3..22801850f339 100644
> > > > > --- a/drivers/iommu/intel/iommu.c
> > > > > +++ b/drivers/iommu/intel/iommu.c
> > > > > @@ -4567,7 +4567,6 @@ static void
> > > > > intel_iommu_get_resv_regions(struct device *device,
> > > > >        }
> > > > >        rcu_read_unlock();
> > > > > 
> > > > > -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
> > > > >        if (dev_is_pci(device)) {
> > > > >            struct pci_dev *pdev = to_pci_dev(device);
> > > > > 
> > > > > @@ -4579,7 +4578,6 @@ static void
> > > > > intel_iommu_get_resv_regions(struct device *device,
> > > > >                    list_add_tail(&reg->list, head);
> > > > >            }
> > > > >        }
> > > > > -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> > > > > 
> > > > >        reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
> > > > >                          IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
> > > > > 
> > > > 
> > > > Hi Baolu,
> > > > 
> > > > I have a question:
> > > > Shouldn't we remove the code between ifdef-endif statements?
> > > > 
> > > > I mean something like this:
> > > > 
> > > > diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
> > > > index b7dff5092fd2..5e077d1c5f5d 100644
> > > > --- a/drivers/iommu/intel/Kconfig
> > > > +++ b/drivers/iommu/intel/Kconfig
> > > > @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
> > > >             to use physical addresses for DMA, at least until this
> > > >             option is removed in the 2.6.32 kernel.
> > > > 
> > > > -config INTEL_IOMMU_FLOPPY_WA
> > > > -       def_bool y
> > > > -       depends on X86
> > > > -       help
> > > > -         Floppy disk drivers are known to bypass DMA API calls
> > > > -         thereby failing to work when IOMMU is enabled. This
> > > > -         workaround will setup a 1:1 mapping for the first
> > > > -         16MiB to make floppy (an ISA device) work.
> > > > -
> > > >    config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
> > > >           bool "Enable Intel IOMMU scalable mode by default"
> > > >           default y
> > > > diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
> > > > index 48cdcd0a5cf3..2c416ad3204e 100644
> > > > --- a/drivers/iommu/intel/iommu.c
> > > > +++ b/drivers/iommu/intel/iommu.c
> > > > @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct
> > > > device *device,
> > > >           }
> > > >           rcu_read_unlock();
> > > > 
> > > > -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
> > > > -       if (dev_is_pci(device)) {
> > > > -               struct pci_dev *pdev = to_pci_dev(device);
> > > > -
> > > > -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
> > > > -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
> > > > -                                       IOMMU_RESV_DIRECT_RELAXABLE,
> > > > -                                       GFP_KERNEL);
> > > > -                       if (reg)
> > > > -                               list_add_tail(&reg->list, head);
> > > > -               }
> > > > -       }
> > > > -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
> > > > -
> > > >           reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
> > > >                                         IOAPIC_RANGE_END -
> > > > IOAPIC_RANGE_START + 1,
> > > >                                         0, IOMMU_RESV_MSI, GFP_KERNEL);
> > > > 
> > > > This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA
> > > > direct mapping region via iommu_get_resv_regions")
> > > 
> > > As long as floppy driver exists in the tree, we have to include above
> > > code. Otherwise, floppy drivers don't work. At least we can easily find
> > > drivers/block/floppy.c which is still maintained (check MAINTAINERS).:-)
> > 
> > But this requires a machine with Intel IOMMU and ISA:
> > 
> > > > > -      16MiB to make floppy (an ISA device) work.
> > 
> > ISA device? I don't believe there are any Intel machines with an IOMMU
> > and an ISA device?
> 
> This workaround was introduced by commit 49a0429e53f2 ("Intel IOMMU:
> Iommu floppy workaround") in 2007. I can't remember what happened 15
> years ago, but I believe there must have been corresponding hardware
> configurations at that time, and the Linux kernel has been maintained it
> to now.

At what point can this be removed then?

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-10 21:00           ` Konrad Rzeszutek Wilk
@ 2022-11-11  2:31             ` Baolu Lu
  2022-11-11 12:36               ` Robin Murphy
  0 siblings, 1 reply; 12+ messages in thread
From: Baolu Lu @ 2022-11-11  2:31 UTC (permalink / raw)
  To: Konrad Rzeszutek Wilk
  Cc: baolu.lu, Harshit Mogalapalli, harshit.m.mogalapalli,
	David Woodhouse, Joerg Roedel, Will Deacon, Robin Murphy, iommu,
	linux-kernel, Darren Kenny

On 2022/11/11 5:00, Konrad Rzeszutek Wilk wrote:
> On Thu, Nov 10, 2022 at 02:39:53PM +0800, Baolu Lu wrote:
>> On 2022/11/10 4:17, Konrad Rzeszutek Wilk wrote:
>>> On Wed, Nov 09, 2022 at 09:16:53PM +0800, Baolu Lu wrote:
>>>> On 2022/11/9 20:16, Harshit Mogalapalli wrote:
>>>>>
>>>>>
>>>>> On 09/11/22 12:35 pm, Baolu Lu wrote:
>>>>>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>>>>>> It is likely that modern intel motherboard will not ship with a
>>>>>>> floppy connection anymore, so let us disable it by default, as it
>>>>>>> gets turned on when we do a make defconfig.
>>>>>>>
>>>>>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>>>>>> Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
>>>>>>> ---
>>>>>>>     drivers/iommu/intel/Kconfig | 2 +-
>>>>>>>     1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>>>
>>>>>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>>>>>> index b7dff5092fd2..c783ae85ca9b 100644
>>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>>           option is removed in the 2.6.32 kernel.
>>>>>>>     config INTEL_IOMMU_FLOPPY_WA
>>>>>>> -    def_bool y
>>>>>>> +    def_bool n
>>>>>>>         depends on X86
>>>>>>>         help
>>>>>>>           Floppy disk drivers are known to bypass DMA API calls
>>>>>>
>>>>>> Nobody selects or depends on this. How about removing this bool? Only
>>>>>> less than 10 lines of code are impacted and are not in any performance
>>>>>> path.
>>>>>>
>>>>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>           to use physical addresses for DMA, at least until this
>>>>>>           option is removed in the 2.6.32 kernel.
>>>>>>
>>>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>>>> -    def_bool y
>>>>>> -    depends on X86
>>>>>> -    help
>>>>>> -      Floppy disk drivers are known to bypass DMA API calls
>>>>>> -      thereby failing to work when IOMMU is enabled. This
>>>>>> -      workaround will setup a 1:1 mapping for the first
>>>>>> -      16MiB to make floppy (an ISA device) work.
>>>>>> -
>>>>>>     config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>>>         bool "Enable Intel IOMMU scalable mode by default"
>>>>>>         default y
>>>>>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>>>>>> index 48cdcd0a5cf3..22801850f339 100644
>>>>>> --- a/drivers/iommu/intel/iommu.c
>>>>>> +++ b/drivers/iommu/intel/iommu.c
>>>>>> @@ -4567,7 +4567,6 @@ static void
>>>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>>>         }
>>>>>>         rcu_read_unlock();
>>>>>>
>>>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>>>         if (dev_is_pci(device)) {
>>>>>>             struct pci_dev *pdev = to_pci_dev(device);
>>>>>>
>>>>>> @@ -4579,7 +4578,6 @@ static void
>>>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>>>                     list_add_tail(&reg->list, head);
>>>>>>             }
>>>>>>         }
>>>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>>>
>>>>>>         reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>>>                           IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>>>>>
>>>>>
>>>>> Hi Baolu,
>>>>>
>>>>> I have a question:
>>>>> Shouldn't we remove the code between ifdef-endif statements?
>>>>>
>>>>> I mean something like this:
>>>>>
>>>>> diff --git a/drivers/iommu/intel/Kconfig b/drivers/iommu/intel/Kconfig
>>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>              to use physical addresses for DMA, at least until this
>>>>>              option is removed in the 2.6.32 kernel.
>>>>>
>>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>>> -       def_bool y
>>>>> -       depends on X86
>>>>> -       help
>>>>> -         Floppy disk drivers are known to bypass DMA API calls
>>>>> -         thereby failing to work when IOMMU is enabled. This
>>>>> -         workaround will setup a 1:1 mapping for the first
>>>>> -         16MiB to make floppy (an ISA device) work.
>>>>> -
>>>>>     config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>>            bool "Enable Intel IOMMU scalable mode by default"
>>>>>            default y
>>>>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>>>>> index 48cdcd0a5cf3..2c416ad3204e 100644
>>>>> --- a/drivers/iommu/intel/iommu.c
>>>>> +++ b/drivers/iommu/intel/iommu.c
>>>>> @@ -4567,20 +4567,6 @@ static void intel_iommu_get_resv_regions(struct
>>>>> device *device,
>>>>>            }
>>>>>            rcu_read_unlock();
>>>>>
>>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>> -       if (dev_is_pci(device)) {
>>>>> -               struct pci_dev *pdev = to_pci_dev(device);
>>>>> -
>>>>> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
>>>>> -                       reg = iommu_alloc_resv_region(0, 1UL << 24, prot,
>>>>> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
>>>>> -                                       GFP_KERNEL);
>>>>> -                       if (reg)
>>>>> -                               list_add_tail(&reg->list, head);
>>>>> -               }
>>>>> -       }
>>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>> -
>>>>>            reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>>                                          IOAPIC_RANGE_END -
>>>>> IOAPIC_RANGE_START + 1,
>>>>>                                          0, IOMMU_RESV_MSI, GFP_KERNEL);
>>>>>
>>>>> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: Expose ISA
>>>>> direct mapping region via iommu_get_resv_regions")
>>>>
>>>> As long as floppy driver exists in the tree, we have to include above
>>>> code. Otherwise, floppy drivers don't work. At least we can easily find
>>>> drivers/block/floppy.c which is still maintained (check MAINTAINERS).:-)
>>>
>>> But this requires a machine with Intel IOMMU and ISA:
>>>
>>>>>> -      16MiB to make floppy (an ISA device) work.
>>>
>>> ISA device? I don't believe there are any Intel machines with an IOMMU
>>> and an ISA device?
>>
>> This workaround was introduced by commit 49a0429e53f2 ("Intel IOMMU:
>> Iommu floppy workaround") in 2007. I can't remember what happened 15
>> years ago, but I believe there must have been corresponding hardware
>> configurations at that time, and the Linux kernel has been maintained it
>> to now.
> 
> At what point can this be removed then?

No floppy block drivers in the tree or all floppy drivers' DMA going
through the kernel DMA APIs.

Best regards,
baolu

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-11  2:31             ` Baolu Lu
@ 2022-11-11 12:36               ` Robin Murphy
  2022-11-11 15:09                 ` Darren Kenny
  0 siblings, 1 reply; 12+ messages in thread
From: Robin Murphy @ 2022-11-11 12:36 UTC (permalink / raw)
  To: Baolu Lu, Konrad Rzeszutek Wilk
  Cc: Harshit Mogalapalli, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, iommu, linux-kernel, Darren Kenny

On 2022-11-11 02:31, Baolu Lu wrote:
> On 2022/11/11 5:00, Konrad Rzeszutek Wilk wrote:
>> On Thu, Nov 10, 2022 at 02:39:53PM +0800, Baolu Lu wrote:
>>> On 2022/11/10 4:17, Konrad Rzeszutek Wilk wrote:
>>>> On Wed, Nov 09, 2022 at 09:16:53PM +0800, Baolu Lu wrote:
>>>>> On 2022/11/9 20:16, Harshit Mogalapalli wrote:
>>>>>>
>>>>>>
>>>>>> On 09/11/22 12:35 pm, Baolu Lu wrote:
>>>>>>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>>>>>>> It is likely that modern intel motherboard will not ship with a
>>>>>>>> floppy connection anymore, so let us disable it by default, as it
>>>>>>>> gets turned on when we do a make defconfig.
>>>>>>>>
>>>>>>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>>>>>>> Signed-off-by: Harshit Mogalapalli 
>>>>>>>> <harshit.m.mogalapalli@oracle.com>
>>>>>>>> ---
>>>>>>>>     drivers/iommu/intel/Kconfig | 2 +-
>>>>>>>>     1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>>>>
>>>>>>>> diff --git a/drivers/iommu/intel/Kconfig 
>>>>>>>> b/drivers/iommu/intel/Kconfig
>>>>>>>> index b7dff5092fd2..c783ae85ca9b 100644
>>>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>>>           option is removed in the 2.6.32 kernel.
>>>>>>>>     config INTEL_IOMMU_FLOPPY_WA
>>>>>>>> -    def_bool y
>>>>>>>> +    def_bool n
>>>>>>>>         depends on X86
>>>>>>>>         help
>>>>>>>>           Floppy disk drivers are known to bypass DMA API calls
>>>>>>>
>>>>>>> Nobody selects or depends on this. How about removing this bool? 
>>>>>>> Only
>>>>>>> less than 10 lines of code are impacted and are not in any 
>>>>>>> performance
>>>>>>> path.
>>>>>>>
>>>>>>> diff --git a/drivers/iommu/intel/Kconfig 
>>>>>>> b/drivers/iommu/intel/Kconfig
>>>>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>>           to use physical addresses for DMA, at least until this
>>>>>>>           option is removed in the 2.6.32 kernel.
>>>>>>>
>>>>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>>>>> -    def_bool y
>>>>>>> -    depends on X86
>>>>>>> -    help
>>>>>>> -      Floppy disk drivers are known to bypass DMA API calls
>>>>>>> -      thereby failing to work when IOMMU is enabled. This
>>>>>>> -      workaround will setup a 1:1 mapping for the first
>>>>>>> -      16MiB to make floppy (an ISA device) work.
>>>>>>> -
>>>>>>>     config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>>>>         bool "Enable Intel IOMMU scalable mode by default"
>>>>>>>         default y
>>>>>>> diff --git a/drivers/iommu/intel/iommu.c 
>>>>>>> b/drivers/iommu/intel/iommu.c
>>>>>>> index 48cdcd0a5cf3..22801850f339 100644
>>>>>>> --- a/drivers/iommu/intel/iommu.c
>>>>>>> +++ b/drivers/iommu/intel/iommu.c
>>>>>>> @@ -4567,7 +4567,6 @@ static void
>>>>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>>>>         }
>>>>>>>         rcu_read_unlock();
>>>>>>>
>>>>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>>>>         if (dev_is_pci(device)) {
>>>>>>>             struct pci_dev *pdev = to_pci_dev(device);
>>>>>>>
>>>>>>> @@ -4579,7 +4578,6 @@ static void
>>>>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>>>>                     list_add_tail(&reg->list, head);
>>>>>>>             }
>>>>>>>         }
>>>>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>>>>
>>>>>>>         reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>>>>                           IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>>>>>>
>>>>>>
>>>>>> Hi Baolu,
>>>>>>
>>>>>> I have a question:
>>>>>> Shouldn't we remove the code between ifdef-endif statements?
>>>>>>
>>>>>> I mean something like this:
>>>>>>
>>>>>> diff --git a/drivers/iommu/intel/Kconfig 
>>>>>> b/drivers/iommu/intel/Kconfig
>>>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>              to use physical addresses for DMA, at least until this
>>>>>>              option is removed in the 2.6.32 kernel.
>>>>>>
>>>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>>>> -       def_bool y
>>>>>> -       depends on X86
>>>>>> -       help
>>>>>> -         Floppy disk drivers are known to bypass DMA API calls
>>>>>> -         thereby failing to work when IOMMU is enabled. This
>>>>>> -         workaround will setup a 1:1 mapping for the first
>>>>>> -         16MiB to make floppy (an ISA device) work.
>>>>>> -
>>>>>>     config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>>>            bool "Enable Intel IOMMU scalable mode by default"
>>>>>>            default y
>>>>>> diff --git a/drivers/iommu/intel/iommu.c 
>>>>>> b/drivers/iommu/intel/iommu.c
>>>>>> index 48cdcd0a5cf3..2c416ad3204e 100644
>>>>>> --- a/drivers/iommu/intel/iommu.c
>>>>>> +++ b/drivers/iommu/intel/iommu.c
>>>>>> @@ -4567,20 +4567,6 @@ static void 
>>>>>> intel_iommu_get_resv_regions(struct
>>>>>> device *device,
>>>>>>            }
>>>>>>            rcu_read_unlock();
>>>>>>
>>>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>>> -       if (dev_is_pci(device)) {
>>>>>> -               struct pci_dev *pdev = to_pci_dev(device);
>>>>>> -
>>>>>> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
>>>>>> -                       reg = iommu_alloc_resv_region(0, 1UL << 
>>>>>> 24, prot,
>>>>>> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
>>>>>> -                                       GFP_KERNEL);
>>>>>> -                       if (reg)
>>>>>> -                               list_add_tail(&reg->list, head);
>>>>>> -               }
>>>>>> -       }
>>>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>>> -
>>>>>>            reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>>>                                          IOAPIC_RANGE_END -
>>>>>> IOAPIC_RANGE_START + 1,
>>>>>>                                          0, IOMMU_RESV_MSI, 
>>>>>> GFP_KERNEL);
>>>>>>
>>>>>> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: 
>>>>>> Expose ISA
>>>>>> direct mapping region via iommu_get_resv_regions")
>>>>>
>>>>> As long as floppy driver exists in the tree, we have to include above
>>>>> code. Otherwise, floppy drivers don't work. At least we can easily 
>>>>> find
>>>>> drivers/block/floppy.c which is still maintained (check 
>>>>> MAINTAINERS).:-)
>>>>
>>>> But this requires a machine with Intel IOMMU and ISA:
>>>>
>>>>>>> -      16MiB to make floppy (an ISA device) work.
>>>>
>>>> ISA device? I don't believe there are any Intel machines with an IOMMU
>>>> and an ISA device?
>>>
>>> This workaround was introduced by commit 49a0429e53f2 ("Intel IOMMU:
>>> Iommu floppy workaround") in 2007. I can't remember what happened 15
>>> years ago, but I believe there must have been corresponding hardware
>>> configurations at that time, and the Linux kernel has been maintained it
>>> to now.
>>
>> At what point can this be removed then?
> 
> No floppy block drivers in the tree or all floppy drivers' DMA going
> through the kernel DMA APIs.

Presumably the point of the config is that ISA bridges aren't expected 
on IA-64, so a tiny code saving can be made there, but at this point is 
anyone really that bothered about any more? There's already tons more 
code all through the driver to support newer features that aren't 
meaningful to old IA-64 hardware, so in my opinion, meh. Given that this 
already won't affect systems that truly can't have any ISA devices, I 
don't see any issue with simply making it unconditional.

Note that plenty of chipsets still have an "ISA" bridge to an LPC 
interface, so ISA DMA is not necessarily as dead as one might like to think.

Robin.

^ permalink raw reply	[flat|nested] 12+ messages in thread
* Re: [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n
  2022-11-11 12:36               ` Robin Murphy
@ 2022-11-11 15:09                 ` Darren Kenny
  0 siblings, 0 replies; 12+ messages in thread
From: Darren Kenny @ 2022-11-11 15:09 UTC (permalink / raw)
  To: Robin Murphy, Baolu Lu, Konrad Rzeszutek Wilk
  Cc: Harshit Mogalapalli, harshit.m.mogalapalli, David Woodhouse,
	Joerg Roedel, Will Deacon, iommu, linux-kernel

On Friday, 2022-11-11 at 12:36:48 UTC, Robin Murphy wrote:
> On 2022-11-11 02:31, Baolu Lu wrote:
>> On 2022/11/11 5:00, Konrad Rzeszutek Wilk wrote:
>>> On Thu, Nov 10, 2022 at 02:39:53PM +0800, Baolu Lu wrote:
>>>> On 2022/11/10 4:17, Konrad Rzeszutek Wilk wrote:
>>>>> On Wed, Nov 09, 2022 at 09:16:53PM +0800, Baolu Lu wrote:
>>>>>> On 2022/11/9 20:16, Harshit Mogalapalli wrote:
>>>>>>>
>>>>>>>
>>>>>>> On 09/11/22 12:35 pm, Baolu Lu wrote:
>>>>>>>> On 2022/11/8 20:58, Harshit Mogalapalli wrote:
>>>>>>>>> It is likely that modern intel motherboard will not ship with a
>>>>>>>>> floppy connection anymore, so let us disable it by default, as it
>>>>>>>>> gets turned on when we do a make defconfig.
>>>>>>>>>
>>>>>>>>> Suggested-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
>>>>>>>>> Signed-off-by: Harshit Mogalapalli 
>>>>>>>>> <harshit.m.mogalapalli@oracle.com>
>>>>>>>>> ---
>>>>>>>>>     drivers/iommu/intel/Kconfig | 2 +-
>>>>>>>>>     1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>>>>>
>>>>>>>>> diff --git a/drivers/iommu/intel/Kconfig 
>>>>>>>>> b/drivers/iommu/intel/Kconfig
>>>>>>>>> index b7dff5092fd2..c783ae85ca9b 100644
>>>>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>>>>> @@ -76,7 +76,7 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>>>>           option is removed in the 2.6.32 kernel.
>>>>>>>>>     config INTEL_IOMMU_FLOPPY_WA
>>>>>>>>> -    def_bool y
>>>>>>>>> +    def_bool n
>>>>>>>>>         depends on X86
>>>>>>>>>         help
>>>>>>>>>           Floppy disk drivers are known to bypass DMA API calls
>>>>>>>>
>>>>>>>> Nobody selects or depends on this. How about removing this bool? 
>>>>>>>> Only
>>>>>>>> less than 10 lines of code are impacted and are not in any 
>>>>>>>> performance
>>>>>>>> path.
>>>>>>>>
>>>>>>>> diff --git a/drivers/iommu/intel/Kconfig 
>>>>>>>> b/drivers/iommu/intel/Kconfig
>>>>>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>>>           to use physical addresses for DMA, at least until this
>>>>>>>>           option is removed in the 2.6.32 kernel.
>>>>>>>>
>>>>>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>>>>>> -    def_bool y
>>>>>>>> -    depends on X86
>>>>>>>> -    help
>>>>>>>> -      Floppy disk drivers are known to bypass DMA API calls
>>>>>>>> -      thereby failing to work when IOMMU is enabled. This
>>>>>>>> -      workaround will setup a 1:1 mapping for the first
>>>>>>>> -      16MiB to make floppy (an ISA device) work.
>>>>>>>> -
>>>>>>>>     config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>>>>>         bool "Enable Intel IOMMU scalable mode by default"
>>>>>>>>         default y
>>>>>>>> diff --git a/drivers/iommu/intel/iommu.c 
>>>>>>>> b/drivers/iommu/intel/iommu.c
>>>>>>>> index 48cdcd0a5cf3..22801850f339 100644
>>>>>>>> --- a/drivers/iommu/intel/iommu.c
>>>>>>>> +++ b/drivers/iommu/intel/iommu.c
>>>>>>>> @@ -4567,7 +4567,6 @@ static void
>>>>>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>>>>>         }
>>>>>>>>         rcu_read_unlock();
>>>>>>>>
>>>>>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>>>>>         if (dev_is_pci(device)) {
>>>>>>>>             struct pci_dev *pdev = to_pci_dev(device);
>>>>>>>>
>>>>>>>> @@ -4579,7 +4578,6 @@ static void
>>>>>>>> intel_iommu_get_resv_regions(struct device *device,
>>>>>>>>                     list_add_tail(&reg->list, head);
>>>>>>>>             }
>>>>>>>>         }
>>>>>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>>>>>
>>>>>>>>         reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>>>>>                           IOAPIC_RANGE_END - IOAPIC_RANGE_START + 1,
>>>>>>>>
>>>>>>>
>>>>>>> Hi Baolu,
>>>>>>>
>>>>>>> I have a question:
>>>>>>> Shouldn't we remove the code between ifdef-endif statements?
>>>>>>>
>>>>>>> I mean something like this:
>>>>>>>
>>>>>>> diff --git a/drivers/iommu/intel/Kconfig 
>>>>>>> b/drivers/iommu/intel/Kconfig
>>>>>>> index b7dff5092fd2..5e077d1c5f5d 100644
>>>>>>> --- a/drivers/iommu/intel/Kconfig
>>>>>>> +++ b/drivers/iommu/intel/Kconfig
>>>>>>> @@ -75,15 +75,6 @@ config INTEL_IOMMU_BROKEN_GFX_WA
>>>>>>>              to use physical addresses for DMA, at least until this
>>>>>>>              option is removed in the 2.6.32 kernel.
>>>>>>>
>>>>>>> -config INTEL_IOMMU_FLOPPY_WA
>>>>>>> -       def_bool y
>>>>>>> -       depends on X86
>>>>>>> -       help
>>>>>>> -         Floppy disk drivers are known to bypass DMA API calls
>>>>>>> -         thereby failing to work when IOMMU is enabled. This
>>>>>>> -         workaround will setup a 1:1 mapping for the first
>>>>>>> -         16MiB to make floppy (an ISA device) work.
>>>>>>> -
>>>>>>>     config INTEL_IOMMU_SCALABLE_MODE_DEFAULT_ON
>>>>>>>            bool "Enable Intel IOMMU scalable mode by default"
>>>>>>>            default y
>>>>>>> diff --git a/drivers/iommu/intel/iommu.c 
>>>>>>> b/drivers/iommu/intel/iommu.c
>>>>>>> index 48cdcd0a5cf3..2c416ad3204e 100644
>>>>>>> --- a/drivers/iommu/intel/iommu.c
>>>>>>> +++ b/drivers/iommu/intel/iommu.c
>>>>>>> @@ -4567,20 +4567,6 @@ static void 
>>>>>>> intel_iommu_get_resv_regions(struct
>>>>>>> device *device,
>>>>>>>            }
>>>>>>>            rcu_read_unlock();
>>>>>>>
>>>>>>> -#ifdef CONFIG_INTEL_IOMMU_FLOPPY_WA
>>>>>>> -       if (dev_is_pci(device)) {
>>>>>>> -               struct pci_dev *pdev = to_pci_dev(device);
>>>>>>> -
>>>>>>> -               if ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) {
>>>>>>> -                       reg = iommu_alloc_resv_region(0, 1UL << 
>>>>>>> 24, prot,
>>>>>>> -                                       IOMMU_RESV_DIRECT_RELAXABLE,
>>>>>>> -                                       GFP_KERNEL);
>>>>>>> -                       if (reg)
>>>>>>> -                               list_add_tail(&reg->list, head);
>>>>>>> -               }
>>>>>>> -       }
>>>>>>> -#endif /* CONFIG_INTEL_IOMMU_FLOPPY_WA */
>>>>>>> -
>>>>>>>            reg = iommu_alloc_resv_region(IOAPIC_RANGE_START,
>>>>>>>                                          IOAPIC_RANGE_END -
>>>>>>> IOAPIC_RANGE_START + 1,
>>>>>>>                                          0, IOMMU_RESV_MSI, 
>>>>>>> GFP_KERNEL);
>>>>>>>
>>>>>>> This code is introduced in Commit d850c2ee5fe2 ("iommu/vt-d: 
>>>>>>> Expose ISA
>>>>>>> direct mapping region via iommu_get_resv_regions")
>>>>>>
>>>>>> As long as floppy driver exists in the tree, we have to include above
>>>>>> code. Otherwise, floppy drivers don't work. At least we can easily 
>>>>>> find
>>>>>> drivers/block/floppy.c which is still maintained (check 
>>>>>> MAINTAINERS).:-)
>>>>>
>>>>> But this requires a machine with Intel IOMMU and ISA:
>>>>>
>>>>>>>> -      16MiB to make floppy (an ISA device) work.
>>>>>
>>>>> ISA device? I don't believe there are any Intel machines with an IOMMU
>>>>> and an ISA device?
>>>>
>>>> This workaround was introduced by commit 49a0429e53f2 ("Intel IOMMU:
>>>> Iommu floppy workaround") in 2007. I can't remember what happened 15
>>>> years ago, but I believe there must have been corresponding hardware
>>>> configurations at that time, and the Linux kernel has been maintained it
>>>> to now.
>>>
>>> At what point can this be removed then?
>> 
>> No floppy block drivers in the tree or all floppy drivers' DMA going
>> through the kernel DMA APIs.
>
> Presumably the point of the config is that ISA bridges aren't expected 
> on IA-64, so a tiny code saving can be made there, but at this point is 
> anyone really that bothered about any more? There's already tons more 
> code all through the driver to support newer features that aren't 
> meaningful to old IA-64 hardware, so in my opinion, meh. Given that this 
> already won't affect systems that truly can't have any ISA devices, I 
> don't see any issue with simply making it unconditional.
>
> Note that plenty of chipsets still have an "ISA" bridge to an LPC 
> interface, so ISA DMA is not necessarily as dead as one might like to think.
>

Based on what you're saying then, the fix that you agree with is what Baolu
first suggested, which is to remove it from Kconfig and just remove the
#ifdef/#endif, but not the code that it contains.

Is that correct?

Thanks,

Darren.

^ permalink raw reply	[flat|nested] 12+ messages in thread
end of thread, other threads:[~2022-11-11 15:22 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-08 12:58 [RFC] iommu/vt-d: set default value of INTEL_IOMMU_FLOPPY_WA to n Harshit Mogalapalli
2022-11-09  7:05 ` Baolu Lu
2022-11-09 12:16   ` Harshit Mogalapalli
2022-11-09 13:16     ` Baolu Lu
2022-11-09 20:17       ` Konrad Rzeszutek Wilk
2022-11-10  6:39         ` Baolu Lu
2022-11-10 21:00           ` Konrad Rzeszutek Wilk
2022-11-11  2:31             ` Baolu Lu
2022-11-11 12:36               ` Robin Murphy
2022-11-11 15:09                 ` Darren Kenny
2022-11-09 13:26     ` Robin Murphy
2022-11-09 13:55       ` Harshit Mogalapalli

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).