Skip to contentSkip to conversation

remap_pfn_range() not suitable for physical RAM?

DISCUSSION Frank Mehnert 2 participants 6 messages Mar 4, 2009 – Mar 14, 2009

lore
Share & tools
lore
#0 lore raw

Hi,

since 2.6.29-rcX it seems to be not allowed anymore to use remap_pfn_range() to remap physical RAM to userland. This restriction applies only if CONFIG_X86_PAT is enabled. It seems that vm_insert_page() does not have this restriction. Am I correct? And what is the reason for this restriction?

Kind regards,

Frank

Signature · 1 line
Dr.-Ing. Frank Mehnert    Sun Microsystems    http://www.sun.com/
#1 lore raw

Hi,

sorry for re-posting this message and sorry for addressing you directly. Perhaps you can help me again or can you give me a pointer for a proper documentation of the various *remap* and *vm_insert* functions:

Quoted 5 lines · Frank Mehnert

since 2.6.29-rcX it seems to be not allowed anymore to use remap_pfn_range() to remap physical RAM to userland. This restriction applies only if CONFIG_X86_PAT is enabled. It seems that vm_insert_page() does not have this restriction. Am I correct? And what is the reason for this restriction?

Kind regards,

Frank

Signature · 1 line
Dr.-Ing. Frank Mehnert    Sun Microsystems    http://www.sun.com/
#2 lore raw

Frank,

Patch here should eliminate the restriction.
http://marc.info/?l=linux-kernel&m=123664923030126&w=1

Thanks,
Venki

Quoted 29 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Frank Mehnert
Sent: Tuesday, March 10, 2009 1:04 AM
To: [email protected]
Cc: [email protected]; Siddha, Suresh B
Subject: Re: remap_pfn_range() not suitable for physical RAM?

Hi,

sorry for re-posting this message and sorry for addressing you directly.
Perhaps you can help me again or can you give me a pointer for a proper documentation of the various *remap* and *vm_insert* functions:

Quoted 3 lines · Frank Mehnert

since 2.6.29-rcX it seems to be not allowed anymore to use remap_pfn_range() to remap physical RAM to userland. This restriction applies only if CONFIG_X86_PAT is enabled. It seems that

vm_insert_page()

Quoted 2 lines

does not have this restriction. Am I correct? And what is the reason for this restriction?

Kind regards,

Frank

Signature · 1 line
Dr.-Ing. Frank Mehnert    Sun Microsystems    http://www.sun.com/
#3 lore raw

Venki,

thanks for your reply. May I ask anyway why this check is done in remap_pfn_range() and vm_insert_pfn() but not in vm_insert_page()?

Kind regards,

Frank

Quoted 44 lines · Venkatesh

Frank,

Patch here should eliminate the restriction.
http://marc.info/?l=linux-kernel&m=123664923030126&w=1

Thanks,
Venki

Quoted 25 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Frank Mehnert
Sent: Tuesday, March 10, 2009 1:04 AM
To: [email protected]
Cc: [email protected]; Siddha, Suresh B
Subject: Re: remap_pfn_range() not suitable for physical RAM?

Hi,

sorry for re-posting this message and sorry for addressing you directly.
Perhaps you can help me again or can you give me a pointer for a proper documentation of the various *remap* and *vm_insert* functions:

Quoted 3 lines · Frank Mehnert

since 2.6.29-rcX it seems to be not allowed anymore to use remap_pfn_range() to remap physical RAM to userland. This restriction applies only if CONFIG_X86_PAT is enabled. It seems that

vm_insert_page()

Quoted 2 lines

does not have this restriction. Am I correct? And what is the reason for this restriction?

Quoted 6 lines

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert Sun Microsystems http://www.sun.com/
--

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 1 line
Dr.-Ing. Frank Mehnert    Sun Microsystems    http://www.sun.com/
#4 lore raw

Frank,

The track and untrack is keeping track of memory type for non-RAM regions and making sure that the type is in sync for various mappings for such addresses. We are not tracking the RAM pages today, so vm_insert_page does not have this. vm_insert_mixed is also not covered today, as we deferred that and tracking of RAM pages into our next set of changes.

Thanks,
Venki

Quoted 57 lines · Frank Mehnert

Venki,

thanks for your reply. May I ask anyway why this check is done in remap_pfn_range() and vm_insert_pfn() but not in vm_insert_page()?

Kind regards,

Frank

Quoted 44 lines · Venkatesh

Frank,

Patch here should eliminate the restriction.
http://marc.info/?l=linux-kernel&m=123664923030126&w=1

Thanks,
Venki

Quoted 25 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Frank Mehnert
Sent: Tuesday, March 10, 2009 1:04 AM
To: [email protected]
Cc: [email protected]; Siddha, Suresh B
Subject: Re: remap_pfn_range() not suitable for physical RAM?

Hi,

sorry for re-posting this message and sorry for addressing you directly.
Perhaps you can help me again or can you give me a pointer for a proper documentation of the various *remap* and *vm_insert* functions:

Quoted 3 lines · Frank Mehnert

since 2.6.29-rcX it seems to be not allowed anymore to use remap_pfn_range() to remap physical RAM to userland. This restriction applies only if CONFIG_X86_PAT is enabled. It seems that

vm_insert_page()

Quoted 2 lines

does not have this restriction. Am I correct? And what is the reason for this restriction?

Quoted 6 lines

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert Sun Microsystems http://www.sun.com/
--

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#5 lore raw

Venki,

thanks for these explanations. So vm_insert_page() is only meant for RAM pages and remap_pfn_range() can be used for any mappings (inclusive I/O memory), is that right?

Thanks,

Frank

Quoted 74 lines · Venkatesh

Frank,

The track and untrack is keeping track of memory type for non-RAM regions and making sure that the type is in sync for various mappings for such addresses. We are not tracking the RAM pages today, so vm_insert_page does not have this. vm_insert_mixed is also not covered today, as we deferred that and tracking of RAM pages into our next set of changes.

Thanks,
Venki

Quoted 55 lines · Frank Mehnert

Venki,

thanks for your reply. May I ask anyway why this check is done in remap_pfn_range() and vm_insert_pfn() but not in vm_insert_page()?

Kind regards,

Frank

Quoted 45 lines · Venkatesh

Frank,

Patch here should eliminate the restriction.
http://marc.info/?l=linux-kernel&m=123664923030126&w=1

Thanks,
Venki

Quoted 26 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Frank Mehnert
Sent: Tuesday, March 10, 2009 1:04 AM
To: [email protected]
Cc: [email protected]; Siddha, Suresh B
Subject: Re: remap_pfn_range() not suitable for physical RAM?

Hi,

sorry for re-posting this message and sorry for addressing you directly.
Perhaps you can help me again or can you give me a pointer for a proper documentation of the various *remap* and *vm_insert* functions:

Quoted 3 lines · Frank Mehnert

since 2.6.29-rcX it seems to be not allowed anymore to use remap_pfn_range() to remap physical RAM to userland. This restriction applies only if CONFIG_X86_PAT is enabled. It seems that

vm_insert_page()

Quoted 2 lines

does not have this restriction. Am I correct? And what is the reason for this restriction?

Quoted 6 lines

Kind regards,

Frank
--
Dr.-Ing. Frank Mehnert Sun Microsystems http://www.sun.com/
--

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 1 line
Dr.-Ing. Frank Mehnert    Sun Microsystems    http://www.sun.com/