Skip to contentSkip to conversation

amba: make amba_bustype constant

PATCH In review Kunwu Chan 3 participants 15 messages Aug 23, 2024 – Sep 5, 2024 Arm primecell bus support

lore
Share & tools
lore

Kunwu Chan

author
#0 lore raw

PATCH 1/1 amba: make amba_bustype constant

From: Kunwu Chan <[email protected]>

Since commit d492cc2573a0 ("driver core: device.h: make struct bus_type a const *"), the driver core can properly handle constant struct bus_type, move the amba_bustype variable to be a constant structure as well, placing it into read-only memory which can not be modified at runtime.

---
 drivers/amba/bus.c       | 2 +-
 include/linux/amba/bus.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/amba/bus.c +1 -1
index 34bc880ca20b..0230c43377c1 100644
@@ -435,7 +435,7 @@ static const struct dev_pm_ops amba_pm = {
  * DMA configuration for platform and AMBA bus is same. So here we reuse
  * platform's DMA config routine.
  */
-struct bus_type amba_bustype = {
+const struct bus_type amba_bustype = {
 	.name		= "amba",
 	.dev_groups	= amba_dev_groups,
 	.match		= amba_match,
include/linux/amba/bus.h +1 -1
index 958a55bcc708..dda2f3ea89cb 100644
@@ -105,7 +105,7 @@ enum amba_vendor {
 	AMBA_VENDOR_LSI = 0xb6,
 };
 
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;
 
 #define to_amba_device(d)	container_of_const(d, struct amba_device, dev)
 
Signature · 1 line
2.41.0
#1 lore raw
Quoted 7 lines · Kunwu Chan

From: Kunwu Chan <[email protected]>

Since commit d492cc2573a0 ("driver core: device.h: make struct bus_type a const *"), the driver core can properly handle constant struct bus_type, move the amba_bustype variable to be a constant structure as well, placing it into read-only memory which can not be modified at runtime.

...

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

P.S. You may look at the PNP bus case (some of the latest patches there).

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#2 lore raw

Thanks for the reply.

Quoted 8 lines · Andy Shevchenko
Quoted 7 lines · Kunwu Chan

From: Kunwu Chan <[email protected]>

Since commit d492cc2573a0 ("driver core: device.h: make struct bus_type a const *"), the driver core can properly handle constant struct bus_type, move the amba_bustype variable to be a constant structure as well, placing it into read-only memory which can not be modified at runtime.

Quoted 6 lines

...

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

Sure, i've find the patch  series (https://lore.kernel.org/all/[email protected]/)

Quoted 2 lines

P.S. You may look at the PNP bus case (some of the latest patches there).

I'll change it as PNP bus in v2.

Signature · 2 lines
Thanks,
  Kunwu.Chan
author
#3 lore raw
In reply to Andy Shevchenko
Quoted 8 lines · Andy Shevchenko
Quoted 7 lines · Kunwu Chan

From: Kunwu Chan <[email protected]>

Since commit d492cc2573a0 ("driver core: device.h: make struct bus_type a const *"), the driver core can properly handle constant struct bus_type, move the amba_bustype variable to be a constant structure as well, placing it into read-only memory which can not be modified at runtime.

Quoted 8 lines

...

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

P.S. You may look at the PNP bus case (some of the latest patches there)

I found it problematic in the process of making changes.

There have some usage of amba_bustype outside the AMBA code ,i.e: https://elixir.bootlin.com/linux/v6.10.4/source/drivers/iommu/iommu.c#L155

So, If we make the amba_bustype inside the AMBA code, the outside cannot access.

If only internal access is allowed, it will compile and report an error, how should I modify it next, and any suggestions?

Signature · 2 lines
Thanks,
  Kunwu.Chan
#4 lore raw
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Fri, Aug 23, 2024 at 02:42:03PM +0800, Kunwu Chan wrote:

...

Quoted 7 lines
Quoted 6 lines

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

P.S. You may look at the PNP bus case (some of the latest patches there)

Quoted 7 lines

I found it problematic in the process of making changes.

There have some usage of amba_bustype outside the AMBA code ,i.e:
https://elixir.bootlin.com/linux/v6.10.4/source/drivers/iommu/iommu.c#L155

So, If we make the amba_bustype inside the AMBA code, the outside cannot access.

Yes, that's the idea.

Quoted 2 lines

If only internal access is allowed, it will compile and report an error, how should I modify it next, and any suggestions?

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch);
2) convert user-by-user (N patches);
3) hide the bus type and make it constant.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#5 lore raw
Quoted 18 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Fri, Aug 23, 2024 at 02:42:03PM +0800, Kunwu Chan wrote:

...

Quoted 5 lines
Quoted 5 lines

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

Quoted 8 lines
Quoted 1 line

P.S. You may look at the PNP bus case (some of the latest patches there)

I found it problematic in the process of making changes.

There have some usage of amba_bustype outside the AMBA code ,i.e:
https://elixir.bootlin.com/linux/v6.10.4/source/drivers/iommu/iommu.c#L155

So, If we make the amba_bustype inside the AMBA code, the outside cannot access.

Quoted 9 lines

Yes, that's the idea.

Quoted 2 lines

If only internal access is allowed, it will compile and report an error, how should I modify it next, and any suggestions?

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch);
2) convert user-by-user (N patches);
3) hide the bus type and make it constant.

Here[1] have many use of  amba_bustype directly outside the drivers/amba

When I try to hide the amba_bustype by move the extern from include/linux/amba to drivers/amba,

we find some errors: "error: use of undeclared identifier amba_bustype".

[1] https://elixir.bootlin.com/linux/v6.10.4/A/ident/amba_bustype

Signature · 2 lines
Thanks,
  Kunwu.Chan
#6 lore raw
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Fri, Aug 23, 2024 at 02:42:03PM +0800, Kunwu Chan wrote:

...

Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 6 lines

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

P.S. You may look at the PNP bus case (some of the latest patches there)

I found it problematic in the process of making changes.

There have some usage of amba_bustype outside the AMBA code ,i.e:
https://elixir.bootlin.com/linux/v6.10.4/source/drivers/iommu/iommu.c#L155

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

So, If we make the amba_bustype inside the AMBA code, the outside cannot access.

Yes, that's the idea.

Quoted 2 lines

If only internal access is allowed, it will compile and report an error, how should I modify it next, and any suggestions?

(1) vvv

Quoted 4 lines
Quoted 4 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch);
2) convert user-by-user (N patches);
3) hide the bus type and make it constant.

(1) ^^^

Quoted 1 line

Here[1] have many use of  amba_bustype directly outside the drivers/amba

Yes.

Quoted 4 lines

When I try to hide the amba_bustype by move the extern from include/linux/amba to drivers/amba,

we find some errors: "error: use of undeclared identifier amba_bustype".

Yes, that's why I put (1) to how solve this as we usually do in the Linux kernel.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#13 lore raw
In reply to Kunwu Chan
Quoted 7 lines · Kunwu Chan

From: Kunwu Chan <[email protected]>

Since commit d492cc2573a0 ("driver core: device.h: make struct bus_type a const *"), the driver core can properly handle constant struct bus_type, move the amba_bustype variable to be a constant structure as well, placing it into read-only memory which can not be modified at runtime.

I'm happy with this, but as AMBA bus maintainer, it needs to go to my patch system. Please see details in my signature. Please also note that I'm limited in terms of my time in front of the screen at the moment, so I'm not very responsive at the moment. Thanks.

Signature · 6 lines
*** please note that I probably will only be occasionally responsive
*** for an unknown period of time due to recent eye surgery making
*** reading quite difficult.

RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
author
#7 lore raw
In reply to Andy Shevchenko

Thanks for the reply.

Quoted 18 lines · Andy Shevchenko
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Fri, Aug 23, 2024 at 02:42:03PM +0800, Kunwu Chan wrote:

...

Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 6 lines

include/linux/amba/bus.h enum amba_vendor { · line 108

Quoted 2 lines
include/linux/amba/bus.h +1 -1
-extern struct bus_type amba_bustype;
+extern const struct bus_type amba_bustype;

Can we actually hide this from the outside, i.e. make it static in the C file, and introduce the dev_is_amba() helper instead?

P.S. You may look at the PNP bus case (some of the latest patches there)

I found it problematic in the process of making changes.

There have some usage of amba_bustype outside the AMBA code ,i.e:
https://elixir.bootlin.com/linux/v6.10.4/source/drivers/iommu/iommu.c#L155

Quoted 10 lines
Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

So, If we make the amba_bustype inside the AMBA code, the outside cannot access.

Yes, that's the idea.

Quoted 2 lines

If only internal access is allowed, it will compile and report an error, how should I modify it next, and any suggestions?

(1) vvv

Quoted 2 lines
Quoted 2 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch)

Done.

Quoted 1 line
Quoted 1 line
Quoted 1 line

2) convert user-by-user (N patches)

I've no idea about how to modify, such as in iommu.c:

staticconststructbus_type <https://elixir.bootlin.com/linux/v6.10.4/C/ident/bus_type>*constiommu_buses <https://elixir.bootlin.com/linux/v6.10.4/C/ident/iommu_buses>[]={
&platform_bus_type <https://elixir.bootlin.com/linux/v6.10.4/C/ident/platform_bus_type>,
#ifdef CONFIG_PCI <https://elixir.bootlin.com/linux/v6.10.4/K/ident/CONFIG_PCI>
&pci_bus_type <https://elixir.bootlin.com/linux/v6.10.4/C/ident/pci_bus_type>,
#endif
#ifdef CONFIG_ARM_AMBA <https://elixir.bootlin.com/linux/v6.10.4/K/ident/CONFIG_ARM_AMBA>
&amba_bustype <https://elixir.bootlin.com/linux/v6.10.4/C/ident/amba_bustype>, Or in arch/arm/mach-highbank/highbank.c#L150 bus_register_notifier <https://elixir.bootlin.com/linux/v6.10.4/C/ident/bus_register_notifier>(&amba_bustype <https://elixir.bootlin.com/linux/v6.10.4/C/ident/amba_bustype>,&highbank_amba_nb <https://elixir.bootlin.com/linux/v6.10.4/C/ident/highbank_amba_nb>);

Quoted 1 line
Quoted 1 line
Quoted 1 line

3) hide the bus type and make it constant.

Done.

Move the 'extern struct' to drivers/amba/bus.c introduce some compilation errors.

Quoted 12 lines

(1) ^^^

Quoted 1 line

Here[1] have many use of  amba_bustype directly outside the drivers/amba

Yes.

Quoted 4 lines

When I try to hide the amba_bustype by move the extern from include/linux/amba to drivers/amba,

we find some errors: "error: use of undeclared identifier amba_bustype".

Yes, that's why I put (1) to how solve this as we usually do in the Linux kernel.

I check the pnp_bus_type code, it did'nt use it the non-PNP code.

So I thought I should add a patch for dev_is_amba as your suggested and just leave it intact and submit it together.

Signature · 2 lines
Thanks,
  Kunwu.Chan
author
#14 lore raw
In reply to Russell King (Oracle)
Quoted 8 lines · Russell King (Oracle)
Quoted 7 lines · Kunwu Chan

From: Kunwu Chan <[email protected]>

Since commit d492cc2573a0 ("driver core: device.h: make struct bus_type a const *"), the driver core can properly handle constant struct bus_type, move the amba_bustype variable to be a constant structure as well, placing it into read-only memory which can not be modified at runtime.

Quoted 5 lines

I'm happy with this, but as AMBA bus maintainer, it needs to go to my patch system. Please see details in my signature. Please also note that I'm limited in terms of my time in front of the screen at the moment, so I'm not very responsive at the moment. Thanks.

Thank you very much for your reply.

 I will follow your suggestion and submit the patch to the https://www.armlinux.org.uk/,

I have already created an account.

Signature · 2 lines
Thanks,
  Kunwu.Chan
#8 lore raw
In reply to Kunwu Chan
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Mon, Aug 26, 2024 at 06:08:11PM +0800, Kunwu Chan wrote:

...

Quoted 6 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch)

Done.

Quoted 1 line
Quoted 1 line
Quoted 1 line

2) convert user-by-user (N patches)

I've no idea about how to modify, such as in iommu.c:

Oh, crap. Yes, this need more thinking.
Anyway, Russell is okay with your initial patch, the rest can be done separately. For now probably we can leave it exported for this only case.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#9 lore raw

Thanks for the reply.

Quoted 7 lines · Andy Shevchenko
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Mon, Aug 26, 2024 at 06:08:11PM +0800, Kunwu Chan wrote:

...

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch)

Done.

Quoted 1 line
Quoted 1 line
Quoted 1 line

2) convert user-by-user (N patches)

I've no idea about how to modify, such as in iommu.c:

Oh, crap. Yes, this need more thinking.
Anyway, Russell is okay with your initial patch, the rest can be done separately. For now probably we can leave it exported for this only case.

Anyway, thanks for the suggestion, i'll   introduce the dev_is_amba function

and  add the Suggested tag for you.

If you have a better idea and need me to do it, you can always contact me.

Signature · 2 lines
Thanks,
  Kunwu.Chan
#10 lore raw
Quoted 18 lines · Kunwu Chan

Thanks for the reply.

Quoted 7 lines · Andy Shevchenko
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Mon, Aug 26, 2024 at 06:08:11PM +0800, Kunwu Chan wrote:

...

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch)

Done.

Quoted 1 line
Quoted 1 line
Quoted 1 line

2) convert user-by-user (N patches)

I've no idea about how to modify, such as in iommu.c:

Oh, crap. Yes, this need more thinking.
Anyway, Russell is okay with your initial patch, the rest can be done separately. For now probably we can leave it exported for this only case.

Quoted 5 lines

Anyway, thanks for the suggestion, i'll   introduce the dev_is_amba function

and  add the Suggested tag for you.

If you have a better idea and need me to do it, you can always contact me.

General policy is not to add stuff that doesn't have any users. From what I can see from briefly reading this discussion, and looking at the patches submitted to me, the dev_is_amba() patch adds a helper, but as yet there are no users - and not even any patches on a mailing list to make use of this helper. Therefore, I won't be applying that patch.

Good idea, but it needs users...

Please note that I likely won't be reading further discussion (see my signature below, and I'm having the same op on the other eye - which is the only eye suitable for screen work at the moment - this Friday.)

Signature · 6 lines
*** please note that I probably will only be occasionally responsive
*** for an unknown period of time due to recent eye surgery making
*** reading quite difficult.

RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!
#11 lore raw
Quoted 8 lines · Russell King (Oracle)
Quoted 7 lines · Kunwu Chan

Thanks for the reply.

Quoted 5 lines · Andy Shevchenko
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Mon, Aug 26, 2024 at 06:08:11PM +0800, Kunwu Chan wrote:

...

Quoted 15 lines
Quoted 9 lines
Quoted 9 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch)

Done.

Quoted 1 line
Quoted 1 line
Quoted 1 line

2) convert user-by-user (N patches)

I've no idea about how to modify, such as in iommu.c:

Oh, crap. Yes, this need more thinking.
Anyway, Russell is okay with your initial patch, the rest can be done separately. For now probably we can leave it exported for this only case.

Quoted 5 lines

Anyway, thanks for the suggestion, i'll   introduce the dev_is_amba function

and  add the Suggested tag for you.

If you have a better idea and need me to do it, you can always contact me.

Quoted 8 lines

General policy is not to add stuff that doesn't have any users. From what I can see from briefly reading this discussion, and looking at the patches submitted to me, the dev_is_amba() patch adds a helper, but as yet there are no users - and not even any patches on a mailing list to make use of this helper. Therefore, I won't be applying that patch.

Good idea, but it needs users...

I agree, we need at least a couple of user to convert.

Quoted 4 lines

Please note that I likely won't be reading further discussion (see my signature below, and I'm having the same op on the other eye - which is the only eye suitable for screen work at the moment - this Friday.)

Oh, wish you a successful op and fast recovery!

Signature · 2 lines
With Best Regards,
Andy Shevchenko
author
#12 lore raw
In reply to Russell King (Oracle)

Thanks for the reply.

Quoted 23 lines · Russell King (Oracle)
Quoted 22 lines · Kunwu Chan

Thanks for the reply.

Quoted 7 lines · Andy Shevchenko
Quoted 4 lines · Kunwu Chan
Quoted 3 lines · Andy Shevchenko
Quoted 2 lines · Kunwu Chan
Quoted 1 line · Andy Shevchenko

On Mon, Aug 26, 2024 at 06:08:11PM +0800, Kunwu Chan wrote:

...

Quoted 9 lines
Quoted 5 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Make it patch series:
1) patch that introduces exported function called dev_is_amba() (1 patch)

Done.

Quoted 1 line
Quoted 1 line
Quoted 1 line

2) convert user-by-user (N patches)

I've no idea about how to modify, such as in iommu.c:

Oh, crap. Yes, this need more thinking.
Anyway, Russell is okay with your initial patch, the rest can be done separately. For now probably we can leave it exported for this only case.

Anyway, thanks for the suggestion, i'll   introduce the dev_is_amba function

and  add the Suggested tag for you.

Quoted 7 lines
Quoted 1 line

If you have a better idea and need me to do it, you can always contact me.

General policy is not to add stuff that doesn't have any users. From what I can see from briefly reading this discussion, and looking at the patches submitted to me, the dev_is_amba() patch adds a helper, but as yet there are no users - and not even any patches on a mailing list to make use of this helper. Therefore, I won't be applying that patch.

Actually, here is a user in [1], i'am ready to use it  in [2].

#ifdef CONFIG_ARM_AMBA
else if (dev->bus == &amba_bustype)

[1] https://elixir.bootlin.com/linux/v6.10.8/source/drivers/of/platform.c#L631

[2] https://lore.kernel.org/all/[email protected]/

Quoted 6 lines

Good idea, but it needs users...

Please note that I likely won't be reading further discussion (see my signature below, and I'm having the same op on the other eye - which is the only eye suitable for screen work at the moment - this Friday.)

Sure, i've send it to the armlinux.org.uk as:

https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=9415/1

https://www.armlinux.org.uk/developer/patches/viewpatch.php?id=9416/1

Good health.

Signature · 2 lines
Thanks,
  Kunwu.Chan