Skip to contentSkip to conversation

FRV: linking error?

QUESTION Cyrill Gorcunov 3 participants 9 messages Feb 18, 2008 – Feb 20, 2008

lore
Share & tools
lore
#0 lore raw

Hi David,

being see over vmlinux.lds for FRV architecture I found the string:

	. = ALIGN(4096);
	.data.page_aligned : { *(.data.idt) }

though the PAGE_SIZE is 16K. Can't figure out why is that... Do you have a few spare minutes to explain, please?

- Cyrill -

#1 lore raw
Quoted 7 lines · Cyrill Gorcunov

being see over vmlinux.lds for FRV architecture I found the string:

	. = ALIGN(4096);
	.data.page_aligned : { *(.data.idt) }

though the PAGE_SIZE is 16K. Can't figure out why is that... Do you have a few spare minutes to explain, please?

Ummm... Good question. Probably a legacy thing. I think the .data.idt section is also an x86-only thing and should be dispensed with for FRV. Leastways, the section has nothing in it according to the link map.

David

#2 lore raw
Quoted 7 lines · David Howells
Quoted 5 lines · Cyrill Gorcunov

being see over vmlinux.lds for FRV architecture I found the string:

	. = ALIGN(4096);
	.data.page_aligned : { *(.data.idt) }
Quoted 6 lines
Quoted 2 lines

though the PAGE_SIZE is 16K. Can't figure out why is that... Do you have a few spare minutes to explain, please?

Ummm... Good question. Probably a legacy thing. I think the .data.idt section is also an x86-only thing and should be dispensed with for FRV. Leastways, the section has nothing in it according to the link map.

A quick grep agree with you.
I see it used only in x86 in traps_32 for the F0 0F bug. It exist in several .lds files - including frv.

Sam

#3 lore raw

[Sam Ravnborg - Wed, Feb 20, 2008 at 02:14:42PM +0100]

| On Wed, Feb 20, 2008 at 01:01:05PM +0000, David Howells wrote:
| > Cyrill Gorcunov <[email protected]> wrote:
| > 
| > > being see over vmlinux.lds for FRV architecture I found the string:
| > > 
| > > 	. = ALIGN(4096);
| > > 	.data.page_aligned : { *(.data.idt) }
| > > 
| > > though the PAGE_SIZE is 16K. Can't figure out why is that...
| > > Do you have a few spare minutes to explain, please?
| > 
| > Ummm...  Good question.  Probably a legacy thing.  I think the .data.idt
| > section is also an x86-only thing and should be dispensed with for FRV.
| > Leastways, the section has nothing in it according to the link map.
| 
| A quick grep agree with you.
| I see it used only in x86 in traps_32 for the F0 0F bug.
| It exist in several .lds files - including frv.
| 
| 	Sam
| 

Sam, maybe we should just eliminate this section at least for FRV?

- Cyrill -

#4 lore raw
Quoted 1 line · Cyrill Gorcunov

Sam, maybe we should just eliminate this section at least for FRV?

You should have a patch in your inbox to do just that.

David

#5 lore raw

[David Howells - Wed, Feb 20, 2008 at 06:13:15PM +0000]

| Cyrill Gorcunov <[email protected]> wrote:
| 
| > Sam, maybe we should just eliminate this section at least for FRV?
| 
| You should have a patch in your inbox to do just that.
| 
| David
| 

Thanks David! I've got them all.
I think Sam will take care of them to be included in mainline. Thanks.

- Cyrill -

#6 lore raw
Quoted 10 lines · Cyrill Gorcunov

[David Howells - Wed, Feb 20, 2008 at 06:13:15PM +0000]

| Cyrill Gorcunov <[email protected]> wrote:
| 
| > Sam, maybe we should just eliminate this section at least for FRV?
| 
| You should have a patch in your inbox to do just that.
| 
| David
| 
Quoted 2 lines

Thanks David! I've got them all.
I think Sam will take care of them to be included in mainline.

arch stuff like this is preferably going via the arch maintainer. Sometimes I do all-arch patches but when they are independent like this one they should go via the arch maintainer.

Cyrill - I would like you to do the same cleanup for m32r + mn10300 + s390
and send the patch to the respective maintainers with a cc to me.

Thanks,
Sam

#7 lore raw

[Sam Ravnborg - Wed, Feb 20, 2008 at 08:19:38PM +0100]

| On Wed, Feb 20, 2008 at 09:15:00PM +0300, Cyrill Gorcunov wrote:
| > [David Howells - Wed, Feb 20, 2008 at 06:13:15PM +0000]
| > | Cyrill Gorcunov <[email protected]> wrote:
| > | 
| > | > Sam, maybe we should just eliminate this section at least for FRV?
| > | 
| > | You should have a patch in your inbox to do just that.
| > | 
| > | David
| > | 
| > 
| > Thanks David! I've got them all.
| > I think Sam will take care of them to be included in mainline.
| 
| arch stuff like this is preferably going via the arch maintainer.
| Sometimes I do all-arch patches but when they are independent
| like this one they should go via the arch maintainer.
| 
| Cyrill - I would like you to do the same cleanup for
| m32r + mn10300 + s390
| and send the patch to the respective maintainers with a cc to me.
| 
| Thanks,
| 	Sam
| 

ok, I will do it

- Cyrill -

#8 lore raw
In reply to David Howells

[David Howells - Wed, Feb 20, 2008 at 01:01:05PM +0000]

| Cyrill Gorcunov <[email protected]> wrote:
| 
| > being see over vmlinux.lds for FRV architecture I found the string:
| > 
| > 	. = ALIGN(4096);
| > 	.data.page_aligned : { *(.data.idt) }
| > 
| > though the PAGE_SIZE is 16K. Can't figure out why is that...
| > Do you have a few spare minutes to explain, please?
| 
| Ummm...  Good question.  Probably a legacy thing.  I think the .data.idt
| section is also an x86-only thing and should be dispensed with for FRV.
| Leastways, the section has nothing in it according to the link map.
| 
| David
| 

Oops ;) Thanks, David

- Cyrill -