mbox series

[v8,0/4] Carry forward IMA measurement log on kexec on ARM64

Message ID 20201030174429.29893-1-nramas@linux.microsoft.com
Headers show
Series Carry forward IMA measurement log on kexec on ARM64 | expand

Message

nramas Oct. 30, 2020, 5:44 p.m. UTC
On kexec file load Integrity Measurement Architecture (IMA) subsystem
may verify the IMA signature of the kernel and initramfs, and measure
it. The command line parameters passed to the kernel in the kexec call
may also be measured by IMA. A remote attestation service can verify
the measurement through the IMA log and the TPM PCR data. This can be
achieved only if the IMA measurement log is carried over from
the current kernel to the next kernel across the kexec call.
However in the current implementation the IMA measurement logs are not
carried over on ARM64 platforms. Therefore a remote attestation service
cannot verify the authenticity of the running kernel on ARM64 platforms
when the kernel is updated through the kexec system call.

This patch series adds support for carrying forward the IMA measurement
log on kexec on ARM64. powerpc already supports carrying forward
the IMA measurement log on kexec.

This series refactors the platform independent code defined for powerpc
such that it can be reused for ARM64 as well. A chosen node namely
"linux,ima-kexec-buffer" is added to the DTB for ARM64 to hold
the address and the size of the memory reserved to carry
the IMA measurement log.

This patch series has been tested for ARM64 platform using QEMU.
I would like help from the community for testing this change on powerpc.
Thanks.

This patch series is based on
commit 598a597636f8 ("Merge tag 'afs-fixes-20201029' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs")
in https://github.com/torvalds/linux "master" branch.

Changelog:

v8:
  - Moved remove_ima_kexec_buffer(), do_get_kexec_buffer(), and
    delete_fdt_mem_rsv() to drivers/of/fdt.c
  - Moved ima_dump_measurement_list() and ima_add_kexec_buffer()
    back to security/integrity/ima/ima_kexec.c

v7:
  - Renamed remove_ima_buffer() to remove_ima_kexec_buffer() and moved
    this function definition to kernel.
  - Moved delete_fdt_mem_rsv() definition to kernel
  - Moved ima_dump_measurement_list() and ima_add_kexec_buffer() to
    a new file namely ima_kexec_fdt.c in IMA

v6:
  - Remove any existing FDT_PROP_IMA_KEXEC_BUFFER property in the device
    tree and also its corresponding memory reservation in the currently
    running kernel.
  - Moved the function remove_ima_buffer() defined for powerpc to IMA
    and renamed the function to ima_remove_kexec_buffer(). Also, moved
    delete_fdt_mem_rsv() from powerpc to IMA.

v5:
  - Merged get_addr_size_cells() and do_get_kexec_buffer() into a single
    function when moving the arch independent code from powerpc to IMA
  - Reverted the change to use FDT functions in powerpc code and added
    back the original code in get_addr_size_cells() and
    do_get_kexec_buffer() for powerpc.
  - Added fdt_add_mem_rsv() for ARM64 to reserve the memory for
    the IMA log buffer during kexec.
  - Fixed the warning reported by kernel test bot for ARM64
    arch_ima_add_kexec_buffer() - moved this function to a new file
    namely arch/arm64/kernel/ima_kexec.c

v4:
  - Submitting the patch series on behalf of the original author
    Prakhar Srivastava <prsriva@linux.microsoft.com>
  - Moved FDT_PROP_IMA_KEXEC_BUFFER ("linux,ima-kexec-buffer") to
    libfdt.h so that it can be shared by multiple platforms.

v3:
Breakup patches further into separate patches.
  - Refactoring non architecture specific code out of powerpc
  - Update powerpc related code to use fdt functions
  - Update IMA buffer read related code to use of functions
  - Add support to store the memory information of the IMA
    measurement logs to be carried forward.
  - Update the property strings to align with documented nodes
    https://github.com/devicetree-org/dt-schema/pull/46

v2:
  Break patches into separate patches.
  - Powerpc related Refactoring
  - Updating the docuemntation for chosen node
  - Updating arm64 to support IMA buffer pass

v1:
  Refactoring carrying over IMA measuremnet logs over Kexec. This patch
    moves the non-architecture specific code out of powerpc and adds to
    security/ima.(Suggested by Thiago)
  Add Documentation regarding the ima-kexec-buffer node in the chosen
    node documentation

v0:
  Add a layer of abstraction to use the memory reserved by device tree
    for ima buffer pass.
  Add support for ima buffer pass using reserved memory for arm64 kexec.
    Update the arch sepcific code path in kexec file load to store the
    ima buffer in the reserved memory. The same reserved memory is read
    on kexec or cold boot.

Lakshmi Ramasubramanian (4):
  powerpc: Refactor kexec functions to move arch independent code to
    drivers/of
  powerpc: Refactor kexec functions to move arch independent code to ima
  arm64: Store IMA log information in kimage used for kexec
  arm64: Add IMA kexec buffer to DTB

 arch/arm64/Kconfig                     |   1 +
 arch/arm64/include/asm/ima.h           |  18 ++++
 arch/arm64/include/asm/kexec.h         |   3 +
 arch/arm64/kernel/Makefile             |   1 +
 arch/arm64/kernel/ima_kexec.c          |  34 ++++++++
 arch/arm64/kernel/machine_kexec_file.c |  18 ++++
 arch/powerpc/include/asm/ima.h         |  13 +--
 arch/powerpc/include/asm/kexec.h       |   1 -
 arch/powerpc/kexec/Makefile            |   7 +-
 arch/powerpc/kexec/file_load.c         |  33 +-------
 arch/powerpc/kexec/ima.c               | 111 ++-----------------------
 drivers/of/fdt.c                       | 110 ++++++++++++++++++++++++
 include/linux/kexec.h                  |  24 ++++++
 include/linux/libfdt.h                 |   3 +
 security/integrity/ima/ima_kexec.c     |  57 +++++++++++++
 15 files changed, 282 insertions(+), 152 deletions(-)
 create mode 100644 arch/arm64/include/asm/ima.h
 create mode 100644 arch/arm64/kernel/ima_kexec.c

Comments

Mimi Zohar Nov. 3, 2020, 2:55 p.m. UTC | #1
Hi Lakshmi,

On Fri, 2020-10-30 at 10:44 -0700, Lakshmi Ramasubramanian wrote:
> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle

> carrying forward the IMA measurement logs on kexec for powerpc do not

> have architecture specific code, but they are currently defined for

> powerpc only.


^ ... logs on kexec, do not have architecture specific code, but are
currently limited to powerpc.

> 

> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove

> the IMA log entry from the device tree and free the memory reserved

> for the log. These functions need to be defined even if the current

> kernel does not support carrying forward IMA log across kexec since

> the previous kernel could have supported that and therefore the current

> kernel needs to free the allocation.


The first paragraph describes these function as "handle carrying
forward the IMA measurement logs on kexec", while this paragraph says
"are used to remove the IMA log entry".  Consider listing all of the
functions being moved in the first paragrah, then "handle carrying
forward" could be expanded to "carrying ... and removing".

> 

> Rename remove_ima_buffer() to remove_ima_kexec_buffer().

> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in

> drivers/of/fdt.c. A later patch in this series will use these functions

> to free the allocation, if any, made by the previous kernel for ARM64.


- ^Define -> Move
- Three functions are being moved, but only two are listed. 
"do_get_kexec_buffer" is not mentioned.
- Don't refer to a later patch, but explain the purpose here.  For
example, "Move ... , making them accessible to other archs."

> 

> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely

> "linux,ima-kexec-buffer", that is added to the DTB to hold

> the address and the size of the memory reserved to carry

> the IMA measurement log.


The above two paragraphs describe renaming a function and defining a
chosen node.  These two preparatory changes should be made,
independently of each other, prior to this patch.  This patch should be
limited to moving code, with the subject line truncated to "move arch
independent code to drivers/of".

thanks,

Mimi
Mimi Zohar Nov. 3, 2020, 3:18 p.m. UTC | #2
On Fri, 2020-10-30 at 10:44 -0700, Lakshmi Ramasubramanian wrote:
> On kexec file load Integrity Measurement Architecture (IMA) subsystem

> may verify the IMA signature of the kernel and initramfs, and measure

> it. The command line parameters passed to the kernel in the kexec call

> may also be measured by IMA. A remote attestation service can verify

> the measurement through the IMA log and the TPM PCR data. This can be

> achieved only if the IMA measurement log is carried over from

> the current kernel to the next kernel across the kexec call.


Nice, but you might want to tweak it a bit.  This is just a suggestion.
"A remote attestation service can verify a TPM quote based on the TPM
event log, the IMA measurement list, and the TPM PCR data".

> However in the current implementation the IMA measurement logs are not

> carried over on ARM64 platforms. Therefore a remote attestation service

> cannot verify the authenticity of the running kernel on ARM64 platforms

> when the kernel is updated through the kexec system call.


The paragraphs above and below are redundant.  The first paragraph
already explained why carrying the measurement across kexec is needed. 
Perhaps drop the above paragraph.

> This patch series adds support for carrying forward the IMA measurement

> log on kexec on ARM64. powerpc already supports carrying forward

> the IMA measurement log on kexec.


And invert these sentences, starting the paragraph with "Powerpc
already" and ending with ARM64.

> 

> This series refactors the platform independent code defined for powerpc

> such that it can be reused for ARM64 as well. A chosen node namely

> "linux,ima-kexec-buffer" is added to the DTB for ARM64 to hold

> the address and the size of the memory reserved to carry

> the IMA measurement log.


^This patch set moves ..."

Thanks,

Mimi

> 

> This patch series has been tested for ARM64 platform using QEMU.

> I would like help from the community for testing this change on powerpc.

> Thanks.

> 

> This patch series is based on

> commit 598a597636f8 ("Merge tag 'afs-fixes-20201029' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs")

> in https://github.com/torvalds/linux "master" branch.
nramas Nov. 3, 2020, 7:15 p.m. UTC | #3
On 11/3/20 6:55 AM, Mimi Zohar wrote:

Hi Mimi,

Thanks for reviewing the patches.

> On Fri, 2020-10-30 at 10:44 -0700, Lakshmi Ramasubramanian wrote:

>> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle

>> carrying forward the IMA measurement logs on kexec for powerpc do not

>> have architecture specific code, but they are currently defined for

>> powerpc only.

> 

> ^ ... logs on kexec, do not have architecture specific code, but are

> currently limited to powerpc.

Will make this change.

> 

>>

>> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove

>> the IMA log entry from the device tree and free the memory reserved

>> for the log. These functions need to be defined even if the current

>> kernel does not support carrying forward IMA log across kexec since

>> the previous kernel could have supported that and therefore the current

>> kernel needs to free the allocation.

> 

> The first paragraph describes these function as "handle carrying

> forward the IMA measurement logs on kexec", while this paragraph says

> "are used to remove the IMA log entry".  Consider listing all of the

> functions being moved in the first paragrah, then "handle carrying

> forward" could be expanded to "carrying ... and removing".

Sure.

> 

>>

>> Rename remove_ima_buffer() to remove_ima_kexec_buffer().

>> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in

>> drivers/of/fdt.c. A later patch in this series will use these functions

>> to free the allocation, if any, made by the previous kernel for ARM64.

> 

> - ^Define -> Move

> - Three functions are being moved, but only two are listed.

> "do_get_kexec_buffer" is not mentioned.

> - Don't refer to a later patch, but explain the purpose here.  For

> example, "Move ... , making them accessible to other archs."

Sure.

> 

>>

>> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely

>> "linux,ima-kexec-buffer", that is added to the DTB to hold

>> the address and the size of the memory reserved to carry

>> the IMA measurement log.

> 

> The above two paragraphs describe renaming a function and defining a

> chosen node.  These two preparatory changes should be made,

> independently of each other, prior to this patch.  This patch should be

> limited to moving code, with the subject line truncated to "move arch

> independent code to drivers/of".

> 


Just to be clear -

Split this patch into 3 parts as listed below:

PATCH #1: Rename remove_ima_buffer() to remove_ima_kexec_buffer()
PATCH #2: Define the chosen node
PATCH #3: Move the functions to drivers/of/fdt.c

Sure - I'll make the above changes and update patch descriptions 
accordingly.

thanks,
  -lakshmi
nramas Nov. 3, 2020, 7:26 p.m. UTC | #4
On 11/3/20 7:18 AM, Mimi Zohar wrote:

Hi Mimi,

> On Fri, 2020-10-30 at 10:44 -0700, Lakshmi Ramasubramanian wrote:

>> On kexec file load Integrity Measurement Architecture (IMA) subsystem

>> may verify the IMA signature of the kernel and initramfs, and measure

>> it. The command line parameters passed to the kernel in the kexec call

>> may also be measured by IMA. A remote attestation service can verify

>> the measurement through the IMA log and the TPM PCR data. This can be

>> achieved only if the IMA measurement log is carried over from

>> the current kernel to the next kernel across the kexec call.

> 

> Nice, but you might want to tweak it a bit.  This is just a suggestion.

> "A remote attestation service can verify a TPM quote based on the TPM

> event log, the IMA measurement list, and the TPM PCR data".

Sure - will make this change.

> 

>> However in the current implementation the IMA measurement logs are not

>> carried over on ARM64 platforms. Therefore a remote attestation service

>> cannot verify the authenticity of the running kernel on ARM64 platforms

>> when the kernel is updated through the kexec system call.

> 

> The paragraphs above and below are redundant.  The first paragraph

> already explained why carrying the measurement across kexec is needed.

> Perhaps drop the above paragraph.

Sure.

> 

>> This patch series adds support for carrying forward the IMA measurement

>> log on kexec on ARM64. powerpc already supports carrying forward

>> the IMA measurement log on kexec.

> 

> And invert these sentences, starting the paragraph with "Powerpc

> already" and ending with ARM64.

Sure.

> 

>>

>> This series refactors the platform independent code defined for powerpc

>> such that it can be reused for ARM64 as well. A chosen node namely

>> "linux,ima-kexec-buffer" is added to the DTB for ARM64 to hold

>> the address and the size of the memory reserved to carry

>> the IMA measurement log.

> 

> ^This patch set moves ..."

Sure - will make this change.

Thanks again for reviewing the patches. Will post the updated patch set 
shortly.

  -lakshmi
Mimi Zohar Nov. 3, 2020, 7:59 p.m. UTC | #5
On Tue, 2020-11-03 at 11:15 -0800, Lakshmi Ramasubramanian wrote:
> On 11/3/20 6:55 AM, Mimi Zohar wrote:

> 

> Hi Mimi,

> 

> Thanks for reviewing the patches.

> 

> > On Fri, 2020-10-30 at 10:44 -0700, Lakshmi Ramasubramanian wrote:

> >> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle

> >> carrying forward the IMA measurement logs on kexec for powerpc do not

> >> have architecture specific code, but they are currently defined for

> >> powerpc only.

> > 

> > ^ ... logs on kexec, do not have architecture specific code, but are

> > currently limited to powerpc.

> Will make this change.

> 

> > 

> >>

> >> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove

> >> the IMA log entry from the device tree and free the memory reserved

> >> for the log. These functions need to be defined even if the current

> >> kernel does not support carrying forward IMA log across kexec since

> >> the previous kernel could have supported that and therefore the current

> >> kernel needs to free the allocation.

> > 

> > The first paragraph describes these function as "handle carrying

> > forward the IMA measurement logs on kexec", while this paragraph says

> > "are used to remove the IMA log entry".  Consider listing all of the

> > functions being moved in the first paragrah, then "handle carrying

> > forward" could be expanded to "carrying ... and removing".

> Sure.


Sorry, even with naming do_get_kexec_buffer(), the measurement list
isn't being carried across kexec.   Please adjust the wording.

> 

> > 

> >>

> >> Rename remove_ima_buffer() to remove_ima_kexec_buffer().

> >> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in

> >> drivers/of/fdt.c. A later patch in this series will use these functions

> >> to free the allocation, if any, made by the previous kernel for ARM64.

> > 

> > - ^Define -> Move

> > - Three functions are being moved, but only two are listed.

> > "do_get_kexec_buffer" is not mentioned.

> > - Don't refer to a later patch, but explain the purpose here.  For

> > example, "Move ... , making them accessible to other archs."

> Sure.

> 

> > 

> >>

> >> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely

> >> "linux,ima-kexec-buffer", that is added to the DTB to hold

> >> the address and the size of the memory reserved to carry

> >> the IMA measurement log.

> > 

> > The above two paragraphs describe renaming a function and defining a

> > chosen node.  These two preparatory changes should be made,

> > independently of each other, prior to this patch.  This patch should be

> > limited to moving code, with the subject line truncated to "move arch

> > independent code to drivers/of".

> > 

> 

> Just to be clear -

> 

> Split this patch into 3 parts as listed below:

> 

> PATCH #1: Rename remove_ima_buffer() to remove_ima_kexec_buffer()

> PATCH #2: Define the chosen node

> PATCH #3: Move the functions to drivers/of/fdt.c


yes, thanks.

Mimi

> 

> Sure - I'll make the above changes and update patch descriptions 

> accordingly.
nramas Nov. 3, 2020, 8:07 p.m. UTC | #6
On 11/3/20 11:59 AM, Mimi Zohar wrote:
> On Tue, 2020-11-03 at 11:15 -0800, Lakshmi Ramasubramanian wrote:

>> On 11/3/20 6:55 AM, Mimi Zohar wrote:

>>

>> Hi Mimi,

>>

>> Thanks for reviewing the patches.

>>

>>> On Fri, 2020-10-30 at 10:44 -0700, Lakshmi Ramasubramanian wrote:

>>>> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle

>>>> carrying forward the IMA measurement logs on kexec for powerpc do not

>>>> have architecture specific code, but they are currently defined for

>>>> powerpc only.

>>>

>>> ^ ... logs on kexec, do not have architecture specific code, but are

>>> currently limited to powerpc.

>> Will make this change.

>>

>>>

>>>>

>>>> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove

>>>> the IMA log entry from the device tree and free the memory reserved

>>>> for the log. These functions need to be defined even if the current

>>>> kernel does not support carrying forward IMA log across kexec since

>>>> the previous kernel could have supported that and therefore the current

>>>> kernel needs to free the allocation.

>>>

>>> The first paragraph describes these function as "handle carrying

>>> forward the IMA measurement logs on kexec", while this paragraph says

>>> "are used to remove the IMA log entry".  Consider listing all of the

>>> functions being moved in the first paragrah, then "handle carrying

>>> forward" could be expanded to "carrying ... and removing".

>> Sure.

> 

> Sorry, even with naming do_get_kexec_buffer(), the measurement list

> isn't being carried across kexec.   Please adjust the wording.

Sure - will update.

> 

>>

>>>

>>>>

>>>> Rename remove_ima_buffer() to remove_ima_kexec_buffer().

>>>> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in

>>>> drivers/of/fdt.c. A later patch in this series will use these functions

>>>> to free the allocation, if any, made by the previous kernel for ARM64.

>>>

>>> - ^Define -> Move

>>> - Three functions are being moved, but only two are listed.

>>> "do_get_kexec_buffer" is not mentioned.

>>> - Don't refer to a later patch, but explain the purpose here.  For

>>> example, "Move ... , making them accessible to other archs."

>> Sure.

>>

>>>

>>>>

>>>> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely

>>>> "linux,ima-kexec-buffer", that is added to the DTB to hold

>>>> the address and the size of the memory reserved to carry

>>>> the IMA measurement log.

>>>

>>> The above two paragraphs describe renaming a function and defining a

>>> chosen node.  These two preparatory changes should be made,

>>> independently of each other, prior to this patch.  This patch should be

>>> limited to moving code, with the subject line truncated to "move arch

>>> independent code to drivers/of".

>>>

>>

>> Just to be clear -

>>

>> Split this patch into 3 parts as listed below:

>>

>> PATCH #1: Rename remove_ima_buffer() to remove_ima_kexec_buffer()

>> PATCH #2: Define the chosen node

>> PATCH #3: Move the functions to drivers/of/fdt.c

> 

> yes, thanks.


thanks,
  -lakshmi
Rob Herring (Arm) Nov. 4, 2020, 10:28 p.m. UTC | #7
On Fri, Oct 30, 2020 at 10:44:26AM -0700, Lakshmi Ramasubramanian wrote:
> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle

> carrying forward the IMA measurement logs on kexec for powerpc do not

> have architecture specific code, but they are currently defined for

> powerpc only.

> 

> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove

> the IMA log entry from the device tree and free the memory reserved

> for the log. These functions need to be defined even if the current

> kernel does not support carrying forward IMA log across kexec since

> the previous kernel could have supported that and therefore the current

> kernel needs to free the allocation.

> 

> Rename remove_ima_buffer() to remove_ima_kexec_buffer().

> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in

> drivers/of/fdt.c. A later patch in this series will use these functions

> to free the allocation, if any, made by the previous kernel for ARM64.

> 

> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely

> "linux,ima-kexec-buffer", that is added to the DTB to hold

> the address and the size of the memory reserved to carry

> the IMA measurement log.

> 

> Co-developed-by: Prakhar Srivastava <prsriva@linux.microsoft.com>

> Signed-off-by: Prakhar Srivastava <prsriva@linux.microsoft.com>

> Signed-off-by: Lakshmi Ramasubramanian <nramas@linux.microsoft.com>

> Reported-by: kernel test robot <lkp@intel.com> error: kernel/kexec_file_fdt.c:30: undefined reference to `fdt_num_mem_rsv'


This should be added for a commit fixing the reported problem. 0-day 
didn't report what this patch implements.

> ---

>  arch/powerpc/include/asm/ima.h   |  10 +--

>  arch/powerpc/include/asm/kexec.h |   1 -

>  arch/powerpc/kexec/file_load.c   |  33 +---------

>  arch/powerpc/kexec/ima.c         |  55 +++-------------

>  drivers/of/fdt.c                 | 110 +++++++++++++++++++++++++++++++

>  include/linux/kexec.h            |  24 +++++++

>  include/linux/libfdt.h           |   3 +

>  7 files changed, 149 insertions(+), 87 deletions(-)

> 

> diff --git a/arch/powerpc/include/asm/ima.h b/arch/powerpc/include/asm/ima.h

> index ead488cf3981..6355a85a3289 100644

> --- a/arch/powerpc/include/asm/ima.h

> +++ b/arch/powerpc/include/asm/ima.h

> @@ -2,17 +2,13 @@

>  #ifndef _ASM_POWERPC_IMA_H

>  #define _ASM_POWERPC_IMA_H

>  

> +#include <linux/kexec.h>

> +

>  struct kimage;

>  

>  int ima_get_kexec_buffer(void **addr, size_t *size);

>  int ima_free_kexec_buffer(void);

>  

> -#ifdef CONFIG_IMA

> -void remove_ima_buffer(void *fdt, int chosen_node);

> -#else

> -static inline void remove_ima_buffer(void *fdt, int chosen_node) {}

> -#endif

> -

>  #ifdef CONFIG_IMA_KEXEC

>  int arch_ima_add_kexec_buffer(struct kimage *image, unsigned long load_addr,

>  			      size_t size);

> @@ -22,7 +18,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node);

>  static inline int setup_ima_buffer(const struct kimage *image, void *fdt,

>  				   int chosen_node)

>  {

> -	remove_ima_buffer(fdt, chosen_node);

> +	remove_ima_kexec_buffer(fdt, chosen_node);

>  	return 0;

>  }

>  #endif /* CONFIG_IMA_KEXEC */

> diff --git a/arch/powerpc/include/asm/kexec.h b/arch/powerpc/include/asm/kexec.h

> index 55d6ede30c19..7c223031ecdd 100644

> --- a/arch/powerpc/include/asm/kexec.h

> +++ b/arch/powerpc/include/asm/kexec.h

> @@ -126,7 +126,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,

>  int setup_new_fdt(const struct kimage *image, void *fdt,

>  		  unsigned long initrd_load_addr, unsigned long initrd_len,

>  		  const char *cmdline);

> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size);

>  

>  #ifdef CONFIG_PPC64

>  struct kexec_buf;

> diff --git a/arch/powerpc/kexec/file_load.c b/arch/powerpc/kexec/file_load.c

> index 9a232bc36c8f..7a17655c530e 100644

> --- a/arch/powerpc/kexec/file_load.c

> +++ b/arch/powerpc/kexec/file_load.c

> @@ -18,6 +18,7 @@

>  #include <linux/kexec.h>

>  #include <linux/of_fdt.h>

>  #include <linux/libfdt.h>

> +#include <linux/kexec.h>

>  #include <asm/setup.h>

>  #include <asm/ima.h>

>  

> @@ -109,38 +110,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,

>  	return 0;

>  }

>  

> -/**

> - * delete_fdt_mem_rsv - delete memory reservation with given address and size

> - *

> - * Return: 0 on success, or negative errno on error.

> - */

> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)

> -{

> -	int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);

> -

> -	for (i = 0; i < num_rsvs; i++) {

> -		uint64_t rsv_start, rsv_size;

> -

> -		ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);

> -		if (ret) {

> -			pr_err("Malformed device tree.\n");

> -			return -EINVAL;

> -		}

> -

> -		if (rsv_start == start && rsv_size == size) {

> -			ret = fdt_del_mem_rsv(fdt, i);

> -			if (ret) {

> -				pr_err("Error deleting device tree reservation.\n");

> -				return -EINVAL;

> -			}

> -

> -			return 0;

> -		}

> -	}

> -

> -	return -ENOENT;

> -}

> -

>  /*

>   * setup_new_fdt - modify /chosen and memory reservation for the next kernel

>   * @image:		kexec image being loaded.

> diff --git a/arch/powerpc/kexec/ima.c b/arch/powerpc/kexec/ima.c

> index 720e50e490b6..2b790230ea15 100644

> --- a/arch/powerpc/kexec/ima.c

> +++ b/arch/powerpc/kexec/ima.c

> @@ -11,6 +11,8 @@

>  #include <linux/of.h>

>  #include <linux/memblock.h>

>  #include <linux/libfdt.h>

> +#include <linux/ima.h>

> +#include <asm/ima.h>

>  

>  static int get_addr_size_cells(int *addr_cells, int *size_cells)

>  {

> @@ -28,24 +30,6 @@ static int get_addr_size_cells(int *addr_cells, int *size_cells)

>  	return 0;

>  }

>  

> -static int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,

> -			       size_t *size)

> -{

> -	int ret, addr_cells, size_cells;

> -

> -	ret = get_addr_size_cells(&addr_cells, &size_cells);

> -	if (ret)

> -		return ret;

> -

> -	if (len < 4 * (addr_cells + size_cells))

> -		return -ENOENT;

> -

> -	*addr = of_read_number(prop, addr_cells);

> -	*size = of_read_number(prop + 4 * addr_cells, size_cells);

> -

> -	return 0;

> -}

> -

>  /**

>   * ima_get_kexec_buffer - get IMA buffer from the previous kernel

>   * @addr:	On successful return, set to point to the buffer contents.

> @@ -100,37 +84,14 @@ int ima_free_kexec_buffer(void)

>  

>  }

>  

> -/**

> - * remove_ima_buffer - remove the IMA buffer property and reservation from @fdt

> - *

> - * The IMA measurement buffer is of no use to a subsequent kernel, so we always

> - * remove it from the device tree.

> - */

> -void remove_ima_buffer(void *fdt, int chosen_node)

> -{

> -	int ret, len;

> -	unsigned long addr;

> -	size_t size;

> -	const void *prop;

> -

> -	prop = fdt_getprop(fdt, chosen_node, "linux,ima-kexec-buffer", &len);

> -	if (!prop)

> -		return;

> -

> -	ret = do_get_kexec_buffer(prop, len, &addr, &size);

> -	fdt_delprop(fdt, chosen_node, "linux,ima-kexec-buffer");

> -	if (ret)

> -		return;

> -

> -	ret = delete_fdt_mem_rsv(fdt, addr, size);

> -	if (!ret)

> -		pr_debug("Removed old IMA buffer reservation.\n");

> -}

> -

>  #ifdef CONFIG_IMA_KEXEC

>  /**

>   * arch_ima_add_kexec_buffer - do arch-specific steps to add the IMA buffer

>   *

> + * @image: kimage struct to set IMA buffer data

> + * @load_addr: Starting address where IMA buffer is loaded at

> + * @size: Number of bytes in the IMA buffer

> + *

>   * Architectures should use this function to pass on the IMA buffer

>   * information to the next kernel.

>   *

> @@ -179,7 +140,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)

>  	int ret, addr_cells, size_cells, entry_size;

>  	u8 value[16];

>  

> -	remove_ima_buffer(fdt, chosen_node);

> +	remove_ima_kexec_buffer(fdt, chosen_node);

>  	if (!image->arch.ima_buffer_size)

>  		return 0;

>  

> @@ -201,7 +162,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)

>  	if (ret)

>  		return ret;

>  

> -	ret = fdt_setprop(fdt, chosen_node, "linux,ima-kexec-buffer", value,

> +	ret = fdt_setprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, value,

>  			  entry_size);

>  	if (ret < 0)

>  		return -EINVAL;

> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c

> index 4602e467ca8b..d2e6f8ce0e42 100644

> --- a/drivers/of/fdt.c

> +++ b/drivers/of/fdt.c

> @@ -25,6 +25,7 @@

>  #include <linux/serial_core.h>

>  #include <linux/sysfs.h>

>  #include <linux/random.h>

> +#include <linux/kexec.h>

>  

>  #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */

>  #include <asm/page.h>

> @@ -1289,4 +1290,113 @@ static int __init of_fdt_raw_init(void)

>  late_initcall(of_fdt_raw_init);

>  #endif

>  

> +#ifdef CONFIG_HAVE_IMA_KEXEC


Can we avoid #ifdef and use IS_ENABLED() within the functions?

> +/**

> + * do_get_kexec_buffer - Get address and size of IMA kexec buffer

> + *

> + * @prop: IMA kexec buffer node in the device tree

> + * @len: Size of the given device tree node property

> + * @addr: Return address of the node

> + * @size: Return size of the node

> + */

> +int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,

> +			size_t *size)

> +{

> +	int addr_cells, size_cells;

> +	struct device_node *root;

> +

> +	root = of_find_node_by_path("/");


The code in fdt.c operates on flat trees. This is an unflattened tree.

> +	if (!root)

> +		return -EINVAL;

> +

> +	addr_cells = of_n_addr_cells(root);

> +	size_cells = of_n_size_cells(root);

> +

> +	of_node_put(root);

> +

> +	if (len < 4 * (addr_cells + size_cells))

> +		return -ENOENT;

> +

> +	*addr = of_read_number(prop, addr_cells);

> +	*size = of_read_number(prop + 4 * addr_cells, size_cells);


There's nothing in this function specific to 'kexec buffer'.

This interface is kind of broken. 'prop' could come from anywhere in 
the tree, but we always read the root address and size cells. Those only 
apply to immediate child node properties. And anything other than 
immediate root child nodes, there needs to be address translation.

> +

> +	return 0;

> +}

> +

> +/**

> + * remove_ima_kexec_buffer - remove the IMA buffer property and

> + *			     reservation from @fdt


IIRC, kerneldoc requires this to be one line.

> + *

> + * @fdt: Flattened Device Tree to update

> + * @chosen_node: Offset to the chosen node in the device tree

> + *

> + * The IMA measurement buffer is of no use to a subsequent kernel,

> + * so we always remove it from the device tree.

> + */

> +void remove_ima_kexec_buffer(void *fdt, int chosen_node)

> +{


Can't this go in some common kexec code?

> +	int ret, len;

> +	unsigned long addr;

> +	size_t size;

> +	const void *prop;

> +

> +	prop = fdt_getprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, &len);

> +	if (!prop) {

> +		pr_debug("Unable to find the ima kexec buffer node\n");

> +		return;

> +	}

> +

> +	ret = do_get_kexec_buffer(prop, len, &addr, &size);

> +	fdt_delprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER);

> +	if (ret) {

> +		pr_err("Unable to delete the ima kexec buffer node\n");

> +		return;

> +	}

> +

> +	ret = delete_fdt_mem_rsv(fdt, addr, size);

> +	if (!ret)

> +		pr_debug("Removed old IMA buffer reservation.\n");

> +}

> +#endif /* CONFIG_HAVE_IMA_KEXEC */

> +

> +#ifdef CONFIG_KEXEC_FILE

> +/**

> + * delete_fdt_mem_rsv - delete memory reservation with given address and size

> + *

> + * @fdt: Flattened Device Tree to update

> + * @start: Starting address of the reservation to delete

> + * @size: Size of the reservation to delete

> + *

> + * Return: 0 on success, or negative errno on error.

> + */

> +int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)

> +{

> +	int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);

> +

> +	for (i = 0; i < num_rsvs; i++) {

> +		uint64_t rsv_start, rsv_size;

> +

> +		ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);

> +		if (ret) {

> +			pr_err("Malformed device tree.\n");

> +			return -EINVAL;

> +		}

> +

> +		if (rsv_start == start && rsv_size == size) {

> +			ret = fdt_del_mem_rsv(fdt, i);

> +			if (ret) {

> +				pr_err("Error deleting device tree reservation.\n");

> +				return -EINVAL;

> +			}

> +

> +			pr_debug("Freed reserved memory at %lu of size %lu\n",

> +				 start, size);

> +			return 0;

> +		}

> +	}

> +

> +	return -ENOENT;

> +}

> +#endif /* CONFIG_KEXEC_FILE */

> +

>  #endif /* CONFIG_OF_EARLY_FLATTREE */

> diff --git a/include/linux/kexec.h b/include/linux/kexec.h

> index 9e93bef52968..6c6c6791a7ba 100644

> --- a/include/linux/kexec.h

> +++ b/include/linux/kexec.h

> @@ -407,6 +407,30 @@ static inline int kexec_crash_loaded(void) { return 0; }

>  #define kexec_in_progress false

>  #endif /* CONFIG_KEXEC_CORE */

>  

> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_HAVE_IMA_KEXEC)


CONFIG_OF_EARLY_FLATTREE is wrong because that's all early boot (i.e. 
init section) functions.

If these functions are implemented in fdt.c, then this is the wrong 
header. But it's the implementation that should move.

> +extern void remove_ima_kexec_buffer(void *fdt, int chosen_node);

> +extern int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,

> +			       size_t *size);

> +#else

> +static inline void remove_ima_kexec_buffer(void *fdt, int chosen_node) {}

> +static inline int do_get_kexec_buffer(const void *prop, int len,

> +				      unsigned long *addr, size_t *size)

> +{

> +	return -EOPNOTSUPP;

> +}

> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_HAVE_IMA_KEXEC */

> +

> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_KEXEC_FILE)

> +extern int delete_fdt_mem_rsv(void *fdt, unsigned long start,

> +			      unsigned long size);

> +#else

> +static inline int delete_fdt_mem_rsv(void *fdt, unsigned long start,

> +				     unsigned long size)

> +{

> +	return 0;

> +}

> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_KEXEC_FILE */

> +

>  #endif /* !defined(__ASSEBMLY__) */

>  

>  #endif /* LINUX_KEXEC_H */

> diff --git a/include/linux/libfdt.h b/include/linux/libfdt.h

> index 90ed4ebfa692..75fb40aa013b 100644

> --- a/include/linux/libfdt.h

> +++ b/include/linux/libfdt.h

> @@ -5,4 +5,7 @@

>  #include <linux/libfdt_env.h>

>  #include "../../scripts/dtc/libfdt/libfdt.h"

>  

> +/* Common device tree properties */

> +#define FDT_PROP_IMA_KEXEC_BUFFER	"linux,ima-kexec-buffer"


This is not part of libfdt. We generally don't do defines for DT 
strings.

> +

>  #endif /* _INCLUDE_LIBFDT_H_ */

> -- 

> 2.29.0

>
nramas Nov. 5, 2020, 12:46 a.m. UTC | #8
On 11/4/20 2:28 PM, Rob Herring wrote:

Hi Rob,

Thanks for reviewing the patch.

> On Fri, Oct 30, 2020 at 10:44:26AM -0700, Lakshmi Ramasubramanian wrote:
>> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle
>> carrying forward the IMA measurement logs on kexec for powerpc do not
>> have architecture specific code, but they are currently defined for
>> powerpc only.
>>
>> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove
>> the IMA log entry from the device tree and free the memory reserved
>> for the log. These functions need to be defined even if the current
>> kernel does not support carrying forward IMA log across kexec since
>> the previous kernel could have supported that and therefore the current
>> kernel needs to free the allocation.
>>
>> Rename remove_ima_buffer() to remove_ima_kexec_buffer().
>> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in
>> drivers/of/fdt.c. A later patch in this series will use these functions
>> to free the allocation, if any, made by the previous kernel for ARM64.
>>
>> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely
>> "linux,ima-kexec-buffer", that is added to the DTB to hold
>> the address and the size of the memory reserved to carry
>> the IMA measurement log.
>>
>> Co-developed-by: Prakhar Srivastava <prsriva@linux.microsoft.com>
>> Signed-off-by: Prakhar Srivastava <prsriva@linux.microsoft.com>
>> Signed-off-by: Lakshmi Ramasubramanian <nramas@linux.microsoft.com>
>> Reported-by: kernel test robot <lkp@intel.com> error: kernel/kexec_file_fdt.c:30: undefined reference to `fdt_num_mem_rsv'
> 
> This should be added for a commit fixing the reported problem. 0-day
> didn't report what this patch implements.
I'll remove that in the updated patch I'll post.

> 
>> ---
>>   arch/powerpc/include/asm/ima.h   |  10 +--
>>   arch/powerpc/include/asm/kexec.h |   1 -
>>   arch/powerpc/kexec/file_load.c   |  33 +---------
>>   arch/powerpc/kexec/ima.c         |  55 +++-------------
>>   drivers/of/fdt.c                 | 110 +++++++++++++++++++++++++++++++
>>   include/linux/kexec.h            |  24 +++++++
>>   include/linux/libfdt.h           |   3 +
>>   7 files changed, 149 insertions(+), 87 deletions(-)
>>
>> diff --git a/arch/powerpc/include/asm/ima.h b/arch/powerpc/include/asm/ima.h
>> index ead488cf3981..6355a85a3289 100644
>> --- a/arch/powerpc/include/asm/ima.h
>> +++ b/arch/powerpc/include/asm/ima.h
>> @@ -2,17 +2,13 @@
>>   #ifndef _ASM_POWERPC_IMA_H
>>   #define _ASM_POWERPC_IMA_H
>>   
>> +#include <linux/kexec.h>
>> +
>>   struct kimage;
>>   
>>   int ima_get_kexec_buffer(void **addr, size_t *size);
>>   int ima_free_kexec_buffer(void);
>>   
>> -#ifdef CONFIG_IMA
>> -void remove_ima_buffer(void *fdt, int chosen_node);
>> -#else
>> -static inline void remove_ima_buffer(void *fdt, int chosen_node) {}
>> -#endif
>> -
>>   #ifdef CONFIG_IMA_KEXEC
>>   int arch_ima_add_kexec_buffer(struct kimage *image, unsigned long load_addr,
>>   			      size_t size);
>> @@ -22,7 +18,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node);
>>   static inline int setup_ima_buffer(const struct kimage *image, void *fdt,
>>   				   int chosen_node)
>>   {
>> -	remove_ima_buffer(fdt, chosen_node);
>> +	remove_ima_kexec_buffer(fdt, chosen_node);
>>   	return 0;
>>   }
>>   #endif /* CONFIG_IMA_KEXEC */
>> diff --git a/arch/powerpc/include/asm/kexec.h b/arch/powerpc/include/asm/kexec.h
>> index 55d6ede30c19..7c223031ecdd 100644
>> --- a/arch/powerpc/include/asm/kexec.h
>> +++ b/arch/powerpc/include/asm/kexec.h
>> @@ -126,7 +126,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,
>>   int setup_new_fdt(const struct kimage *image, void *fdt,
>>   		  unsigned long initrd_load_addr, unsigned long initrd_len,
>>   		  const char *cmdline);
>> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size);
>>   
>>   #ifdef CONFIG_PPC64
>>   struct kexec_buf;
>> diff --git a/arch/powerpc/kexec/file_load.c b/arch/powerpc/kexec/file_load.c
>> index 9a232bc36c8f..7a17655c530e 100644
>> --- a/arch/powerpc/kexec/file_load.c
>> +++ b/arch/powerpc/kexec/file_load.c
>> @@ -18,6 +18,7 @@
>>   #include <linux/kexec.h>
>>   #include <linux/of_fdt.h>
>>   #include <linux/libfdt.h>
>> +#include <linux/kexec.h>
>>   #include <asm/setup.h>
>>   #include <asm/ima.h>
>>   
>> @@ -109,38 +110,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,
>>   	return 0;
>>   }
>>   
>> -/**
>> - * delete_fdt_mem_rsv - delete memory reservation with given address and size
>> - *
>> - * Return: 0 on success, or negative errno on error.
>> - */
>> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)
>> -{
>> -	int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);
>> -
>> -	for (i = 0; i < num_rsvs; i++) {
>> -		uint64_t rsv_start, rsv_size;
>> -
>> -		ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);
>> -		if (ret) {
>> -			pr_err("Malformed device tree.\n");
>> -			return -EINVAL;
>> -		}
>> -
>> -		if (rsv_start == start && rsv_size == size) {
>> -			ret = fdt_del_mem_rsv(fdt, i);
>> -			if (ret) {
>> -				pr_err("Error deleting device tree reservation.\n");
>> -				return -EINVAL;
>> -			}
>> -
>> -			return 0;
>> -		}
>> -	}
>> -
>> -	return -ENOENT;
>> -}
>> -
>>   /*
>>    * setup_new_fdt - modify /chosen and memory reservation for the next kernel
>>    * @image:		kexec image being loaded.
>> diff --git a/arch/powerpc/kexec/ima.c b/arch/powerpc/kexec/ima.c
>> index 720e50e490b6..2b790230ea15 100644
>> --- a/arch/powerpc/kexec/ima.c
>> +++ b/arch/powerpc/kexec/ima.c
>> @@ -11,6 +11,8 @@
>>   #include <linux/of.h>
>>   #include <linux/memblock.h>
>>   #include <linux/libfdt.h>
>> +#include <linux/ima.h>
>> +#include <asm/ima.h>
>>   
>>   static int get_addr_size_cells(int *addr_cells, int *size_cells)
>>   {
>> @@ -28,24 +30,6 @@ static int get_addr_size_cells(int *addr_cells, int *size_cells)
>>   	return 0;
>>   }
>>   
>> -static int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
>> -			       size_t *size)
>> -{
>> -	int ret, addr_cells, size_cells;
>> -
>> -	ret = get_addr_size_cells(&addr_cells, &size_cells);
>> -	if (ret)
>> -		return ret;
>> -
>> -	if (len < 4 * (addr_cells + size_cells))
>> -		return -ENOENT;
>> -
>> -	*addr = of_read_number(prop, addr_cells);
>> -	*size = of_read_number(prop + 4 * addr_cells, size_cells);
>> -
>> -	return 0;
>> -}
>> -
>>   /**
>>    * ima_get_kexec_buffer - get IMA buffer from the previous kernel
>>    * @addr:	On successful return, set to point to the buffer contents.
>> @@ -100,37 +84,14 @@ int ima_free_kexec_buffer(void)
>>   
>>   }
>>   
>> -/**
>> - * remove_ima_buffer - remove the IMA buffer property and reservation from @fdt
>> - *
>> - * The IMA measurement buffer is of no use to a subsequent kernel, so we always
>> - * remove it from the device tree.
>> - */
>> -void remove_ima_buffer(void *fdt, int chosen_node)
>> -{
>> -	int ret, len;
>> -	unsigned long addr;
>> -	size_t size;
>> -	const void *prop;
>> -
>> -	prop = fdt_getprop(fdt, chosen_node, "linux,ima-kexec-buffer", &len);
>> -	if (!prop)
>> -		return;
>> -
>> -	ret = do_get_kexec_buffer(prop, len, &addr, &size);
>> -	fdt_delprop(fdt, chosen_node, "linux,ima-kexec-buffer");
>> -	if (ret)
>> -		return;
>> -
>> -	ret = delete_fdt_mem_rsv(fdt, addr, size);
>> -	if (!ret)
>> -		pr_debug("Removed old IMA buffer reservation.\n");
>> -}
>> -
>>   #ifdef CONFIG_IMA_KEXEC
>>   /**
>>    * arch_ima_add_kexec_buffer - do arch-specific steps to add the IMA buffer
>>    *
>> + * @image: kimage struct to set IMA buffer data
>> + * @load_addr: Starting address where IMA buffer is loaded at
>> + * @size: Number of bytes in the IMA buffer
>> + *
>>    * Architectures should use this function to pass on the IMA buffer
>>    * information to the next kernel.
>>    *
>> @@ -179,7 +140,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)
>>   	int ret, addr_cells, size_cells, entry_size;
>>   	u8 value[16];
>>   
>> -	remove_ima_buffer(fdt, chosen_node);
>> +	remove_ima_kexec_buffer(fdt, chosen_node);
>>   	if (!image->arch.ima_buffer_size)
>>   		return 0;
>>   
>> @@ -201,7 +162,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)
>>   	if (ret)
>>   		return ret;
>>   
>> -	ret = fdt_setprop(fdt, chosen_node, "linux,ima-kexec-buffer", value,
>> +	ret = fdt_setprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, value,
>>   			  entry_size);
>>   	if (ret < 0)
>>   		return -EINVAL;
>> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
>> index 4602e467ca8b..d2e6f8ce0e42 100644
>> --- a/drivers/of/fdt.c
>> +++ b/drivers/of/fdt.c
>> @@ -25,6 +25,7 @@
>>   #include <linux/serial_core.h>
>>   #include <linux/sysfs.h>
>>   #include <linux/random.h>
>> +#include <linux/kexec.h>
>>   
>>   #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */
>>   #include <asm/page.h>
>> @@ -1289,4 +1290,113 @@ static int __init of_fdt_raw_init(void)
>>   late_initcall(of_fdt_raw_init);
>>   #endif
>>   
>> +#ifdef CONFIG_HAVE_IMA_KEXEC
> 
> Can we avoid #ifdef and use IS_ENABLED() within the functions?
I can use IS_ENABLED() or move the functions to another C file and 
conditionally build based on the CONFIG.

> 
>> +/**
>> + * do_get_kexec_buffer - Get address and size of IMA kexec buffer
>> + *
>> + * @prop: IMA kexec buffer node in the device tree
>> + * @len: Size of the given device tree node property
>> + * @addr: Return address of the node
>> + * @size: Return size of the node
>> + */
>> +int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
>> +			size_t *size)
>> +{
>> +	int addr_cells, size_cells;
>> +	struct device_node *root;
>> +
>> +	root = of_find_node_by_path("/");
> 
> The code in fdt.c operates on flat trees. This is an unflattened tree.

Would it be better if I move these functions to a new C file under 
"drivers/of"?

> 
>> +	if (!root)
>> +		return -EINVAL;
>> +
>> +	addr_cells = of_n_addr_cells(root);
>> +	size_cells = of_n_size_cells(root);
>> +
>> +	of_node_put(root);
>> +
>> +	if (len < 4 * (addr_cells + size_cells))
>> +		return -ENOENT;
>> +
>> +	*addr = of_read_number(prop, addr_cells);
>> +	*size = of_read_number(prop + 4 * addr_cells, size_cells);
> 
> There's nothing in this function specific to 'kexec buffer'.
> 
> This interface is kind of broken. 'prop' could come from anywhere in
> the tree, but we always read the root address and size cells. Those only
> apply to immediate child node properties. And anything other than
> immediate root child nodes, there needs to be address translation.
> 

I moved this function from arch/powerpc/kexec/ima.c to here so that it 
can be shared by ARM64 as well. Below is one of the usage of 
do_get_kexec_buffer().

         prop = of_get_property(of_chosen, "linux,ima-kexec-buffer", &len);
         if (!prop)
                 return -ENOENT;

         ret = do_get_kexec_buffer(prop, len, &tmp_addr, &tmp_size);

Is there a way to detect if the 'prop' is an immediate root child node 
or not so that we can handle it appropriately in this interface function?

>> +
>> +	return 0;
>> +}
>> +
>> +/**
>> + * remove_ima_kexec_buffer - remove the IMA buffer property and
>> + *			     reservation from @fdt
> 
> IIRC, kerneldoc requires this to be one line.
I'll upate.

> 
>> + *
>> + * @fdt: Flattened Device Tree to update
>> + * @chosen_node: Offset to the chosen node in the device tree
>> + *
>> + * The IMA measurement buffer is of no use to a subsequent kernel,
>> + * so we always remove it from the device tree.
>> + */
>> +void remove_ima_kexec_buffer(void *fdt, int chosen_node)
>> +{
> 
> Can't this go in some common kexec code?

I'd considered moving this to "kernel", but Thiago suggested 
"drivers/of" is a better place. If you have suggestions for a better 
place, please let me know.

Note that this function needs to be available even when CONFIG_IMA is 
not enabled. So I can't move it to "security/integrity/ima".

> 
>> +	int ret, len;
>> +	unsigned long addr;
>> +	size_t size;
>> +	const void *prop;
>> +
>> +	prop = fdt_getprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, &len);
>> +	if (!prop) {
>> +		pr_debug("Unable to find the ima kexec buffer node\n");
>> +		return;
>> +	}
>> +
>> +	ret = do_get_kexec_buffer(prop, len, &addr, &size);
>> +	fdt_delprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER);
>> +	if (ret) {
>> +		pr_err("Unable to delete the ima kexec buffer node\n");
>> +		return;
>> +	}
>> +
>> +	ret = delete_fdt_mem_rsv(fdt, addr, size);
>> +	if (!ret)
>> +		pr_debug("Removed old IMA buffer reservation.\n");
>> +}
>> +#endif /* CONFIG_HAVE_IMA_KEXEC */
>> +
>> +#ifdef CONFIG_KEXEC_FILE
>> +/**
>> + * delete_fdt_mem_rsv - delete memory reservation with given address and size
>> + *
>> + * @fdt: Flattened Device Tree to update
>> + * @start: Starting address of the reservation to delete
>> + * @size: Size of the reservation to delete
>> + *
>> + * Return: 0 on success, or negative errno on error.
>> + */
>> +int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)
>> +{
>> +	int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);
>> +
>> +	for (i = 0; i < num_rsvs; i++) {
>> +		uint64_t rsv_start, rsv_size;
>> +
>> +		ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);
>> +		if (ret) {
>> +			pr_err("Malformed device tree.\n");
>> +			return -EINVAL;
>> +		}
>> +
>> +		if (rsv_start == start && rsv_size == size) {
>> +			ret = fdt_del_mem_rsv(fdt, i);
>> +			if (ret) {
>> +				pr_err("Error deleting device tree reservation.\n");
>> +				return -EINVAL;
>> +			}
>> +
>> +			pr_debug("Freed reserved memory at %lu of size %lu\n",
>> +				 start, size);
>> +			return 0;
>> +		}
>> +	}
>> +
>> +	return -ENOENT;
>> +}
>> +#endif /* CONFIG_KEXEC_FILE */
>> +
>>   #endif /* CONFIG_OF_EARLY_FLATTREE */
>> diff --git a/include/linux/kexec.h b/include/linux/kexec.h
>> index 9e93bef52968..6c6c6791a7ba 100644
>> --- a/include/linux/kexec.h
>> +++ b/include/linux/kexec.h
>> @@ -407,6 +407,30 @@ static inline int kexec_crash_loaded(void) { return 0; }
>>   #define kexec_in_progress false
>>   #endif /* CONFIG_KEXEC_CORE */
>>   
>> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_HAVE_IMA_KEXEC)
> 
> CONFIG_OF_EARLY_FLATTREE is wrong because that's all early boot (i.e.
> init section) functions.
> 
> If these functions are implemented in fdt.c, then this is the wrong
> header. But it's the implementation that should move.

If you think defining these functions in "drivers/of" is okay, I'll move 
them to a new C file in OF.

Would CONFIG_OF_FLATTREE be a better one to enable these ima kexec 
functions? "of_" and "fdt_" are the functions called by these ima kexec 
functions.

> 
>> +extern void remove_ima_kexec_buffer(void *fdt, int chosen_node);
>> +extern int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
>> +			       size_t *size);
>> +#else
>> +static inline void remove_ima_kexec_buffer(void *fdt, int chosen_node) {}
>> +static inline int do_get_kexec_buffer(const void *prop, int len,
>> +				      unsigned long *addr, size_t *size)
>> +{
>> +	return -EOPNOTSUPP;
>> +}
>> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_HAVE_IMA_KEXEC */
>> +
>> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_KEXEC_FILE)
>> +extern int delete_fdt_mem_rsv(void *fdt, unsigned long start,
>> +			      unsigned long size);
>> +#else
>> +static inline int delete_fdt_mem_rsv(void *fdt, unsigned long start,
>> +				     unsigned long size)
>> +{
>> +	return 0;
>> +}
>> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_KEXEC_FILE */
>> +
>>   #endif /* !defined(__ASSEBMLY__) */
>>   
>>   #endif /* LINUX_KEXEC_H */
>> diff --git a/include/linux/libfdt.h b/include/linux/libfdt.h
>> index 90ed4ebfa692..75fb40aa013b 100644
>> --- a/include/linux/libfdt.h
>> +++ b/include/linux/libfdt.h
>> @@ -5,4 +5,7 @@
>>   #include <linux/libfdt_env.h>
>>   #include "../../scripts/dtc/libfdt/libfdt.h"
>>   
>> +/* Common device tree properties */
>> +#define FDT_PROP_IMA_KEXEC_BUFFER	"linux,ima-kexec-buffer"
> 
> This is not part of libfdt. We generally don't do defines for DT
> strings.
Should I create a new header file for this or would you suggest using an 
existing one?

thanks,
  -lakshmi

> 
>> +
>>   #endif /* _INCLUDE_LIBFDT_H_ */
>> -- 
>> 2.29.0
>>
Rob Herring (Arm) Nov. 5, 2020, 2:38 p.m. UTC | #9
On Wed, Nov 4, 2020 at 6:46 PM Lakshmi Ramasubramanian
<nramas@linux.microsoft.com> wrote:
>
> On 11/4/20 2:28 PM, Rob Herring wrote:
>
> Hi Rob,
>
> Thanks for reviewing the patch.
>
> > On Fri, Oct 30, 2020 at 10:44:26AM -0700, Lakshmi Ramasubramanian wrote:
> >> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle
> >> carrying forward the IMA measurement logs on kexec for powerpc do not
> >> have architecture specific code, but they are currently defined for
> >> powerpc only.
> >>
> >> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove
> >> the IMA log entry from the device tree and free the memory reserved
> >> for the log. These functions need to be defined even if the current
> >> kernel does not support carrying forward IMA log across kexec since
> >> the previous kernel could have supported that and therefore the current
> >> kernel needs to free the allocation.
> >>
> >> Rename remove_ima_buffer() to remove_ima_kexec_buffer().
> >> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in
> >> drivers/of/fdt.c. A later patch in this series will use these functions
> >> to free the allocation, if any, made by the previous kernel for ARM64.
> >>
> >> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely
> >> "linux,ima-kexec-buffer", that is added to the DTB to hold
> >> the address and the size of the memory reserved to carry
> >> the IMA measurement log.
> >>
> >> Co-developed-by: Prakhar Srivastava <prsriva@linux.microsoft.com>
> >> Signed-off-by: Prakhar Srivastava <prsriva@linux.microsoft.com>
> >> Signed-off-by: Lakshmi Ramasubramanian <nramas@linux.microsoft.com>
> >> Reported-by: kernel test robot <lkp@intel.com> error: kernel/kexec_file_fdt.c:30: undefined reference to `fdt_num_mem_rsv'
> >
> > This should be added for a commit fixing the reported problem. 0-day
> > didn't report what this patch implements.
> I'll remove that in the updated patch I'll post.
>
> >
> >> ---
> >>   arch/powerpc/include/asm/ima.h   |  10 +--
> >>   arch/powerpc/include/asm/kexec.h |   1 -
> >>   arch/powerpc/kexec/file_load.c   |  33 +---------
> >>   arch/powerpc/kexec/ima.c         |  55 +++-------------
> >>   drivers/of/fdt.c                 | 110 +++++++++++++++++++++++++++++++
> >>   include/linux/kexec.h            |  24 +++++++
> >>   include/linux/libfdt.h           |   3 +
> >>   7 files changed, 149 insertions(+), 87 deletions(-)
> >>
> >> diff --git a/arch/powerpc/include/asm/ima.h b/arch/powerpc/include/asm/ima.h
> >> index ead488cf3981..6355a85a3289 100644
> >> --- a/arch/powerpc/include/asm/ima.h
> >> +++ b/arch/powerpc/include/asm/ima.h
> >> @@ -2,17 +2,13 @@
> >>   #ifndef _ASM_POWERPC_IMA_H
> >>   #define _ASM_POWERPC_IMA_H
> >>
> >> +#include <linux/kexec.h>
> >> +
> >>   struct kimage;
> >>
> >>   int ima_get_kexec_buffer(void **addr, size_t *size);
> >>   int ima_free_kexec_buffer(void);
> >>
> >> -#ifdef CONFIG_IMA
> >> -void remove_ima_buffer(void *fdt, int chosen_node);
> >> -#else
> >> -static inline void remove_ima_buffer(void *fdt, int chosen_node) {}
> >> -#endif
> >> -
> >>   #ifdef CONFIG_IMA_KEXEC
> >>   int arch_ima_add_kexec_buffer(struct kimage *image, unsigned long load_addr,
> >>                            size_t size);
> >> @@ -22,7 +18,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node);
> >>   static inline int setup_ima_buffer(const struct kimage *image, void *fdt,
> >>                                 int chosen_node)
> >>   {
> >> -    remove_ima_buffer(fdt, chosen_node);
> >> +    remove_ima_kexec_buffer(fdt, chosen_node);
> >>      return 0;
> >>   }
> >>   #endif /* CONFIG_IMA_KEXEC */
> >> diff --git a/arch/powerpc/include/asm/kexec.h b/arch/powerpc/include/asm/kexec.h
> >> index 55d6ede30c19..7c223031ecdd 100644
> >> --- a/arch/powerpc/include/asm/kexec.h
> >> +++ b/arch/powerpc/include/asm/kexec.h
> >> @@ -126,7 +126,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,
> >>   int setup_new_fdt(const struct kimage *image, void *fdt,
> >>                unsigned long initrd_load_addr, unsigned long initrd_len,
> >>                const char *cmdline);
> >> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size);
> >>
> >>   #ifdef CONFIG_PPC64
> >>   struct kexec_buf;
> >> diff --git a/arch/powerpc/kexec/file_load.c b/arch/powerpc/kexec/file_load.c
> >> index 9a232bc36c8f..7a17655c530e 100644
> >> --- a/arch/powerpc/kexec/file_load.c
> >> +++ b/arch/powerpc/kexec/file_load.c
> >> @@ -18,6 +18,7 @@
> >>   #include <linux/kexec.h>
> >>   #include <linux/of_fdt.h>
> >>   #include <linux/libfdt.h>
> >> +#include <linux/kexec.h>
> >>   #include <asm/setup.h>
> >>   #include <asm/ima.h>
> >>
> >> @@ -109,38 +110,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,
> >>      return 0;
> >>   }
> >>
> >> -/**
> >> - * delete_fdt_mem_rsv - delete memory reservation with given address and size
> >> - *
> >> - * Return: 0 on success, or negative errno on error.
> >> - */
> >> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)
> >> -{
> >> -    int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);
> >> -
> >> -    for (i = 0; i < num_rsvs; i++) {
> >> -            uint64_t rsv_start, rsv_size;
> >> -
> >> -            ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);
> >> -            if (ret) {
> >> -                    pr_err("Malformed device tree.\n");
> >> -                    return -EINVAL;
> >> -            }
> >> -
> >> -            if (rsv_start == start && rsv_size == size) {
> >> -                    ret = fdt_del_mem_rsv(fdt, i);
> >> -                    if (ret) {
> >> -                            pr_err("Error deleting device tree reservation.\n");
> >> -                            return -EINVAL;
> >> -                    }
> >> -
> >> -                    return 0;
> >> -            }
> >> -    }
> >> -
> >> -    return -ENOENT;
> >> -}
> >> -
> >>   /*
> >>    * setup_new_fdt - modify /chosen and memory reservation for the next kernel
> >>    * @image:         kexec image being loaded.
> >> diff --git a/arch/powerpc/kexec/ima.c b/arch/powerpc/kexec/ima.c
> >> index 720e50e490b6..2b790230ea15 100644
> >> --- a/arch/powerpc/kexec/ima.c
> >> +++ b/arch/powerpc/kexec/ima.c
> >> @@ -11,6 +11,8 @@
> >>   #include <linux/of.h>
> >>   #include <linux/memblock.h>
> >>   #include <linux/libfdt.h>
> >> +#include <linux/ima.h>
> >> +#include <asm/ima.h>
> >>
> >>   static int get_addr_size_cells(int *addr_cells, int *size_cells)
> >>   {
> >> @@ -28,24 +30,6 @@ static int get_addr_size_cells(int *addr_cells, int *size_cells)
> >>      return 0;
> >>   }
> >>
> >> -static int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
> >> -                           size_t *size)
> >> -{
> >> -    int ret, addr_cells, size_cells;
> >> -
> >> -    ret = get_addr_size_cells(&addr_cells, &size_cells);
> >> -    if (ret)
> >> -            return ret;
> >> -
> >> -    if (len < 4 * (addr_cells + size_cells))
> >> -            return -ENOENT;
> >> -
> >> -    *addr = of_read_number(prop, addr_cells);
> >> -    *size = of_read_number(prop + 4 * addr_cells, size_cells);
> >> -
> >> -    return 0;
> >> -}
> >> -
> >>   /**
> >>    * ima_get_kexec_buffer - get IMA buffer from the previous kernel
> >>    * @addr:  On successful return, set to point to the buffer contents.
> >> @@ -100,37 +84,14 @@ int ima_free_kexec_buffer(void)
> >>
> >>   }
> >>
> >> -/**
> >> - * remove_ima_buffer - remove the IMA buffer property and reservation from @fdt
> >> - *
> >> - * The IMA measurement buffer is of no use to a subsequent kernel, so we always
> >> - * remove it from the device tree.
> >> - */
> >> -void remove_ima_buffer(void *fdt, int chosen_node)
> >> -{
> >> -    int ret, len;
> >> -    unsigned long addr;
> >> -    size_t size;
> >> -    const void *prop;
> >> -
> >> -    prop = fdt_getprop(fdt, chosen_node, "linux,ima-kexec-buffer", &len);
> >> -    if (!prop)
> >> -            return;
> >> -
> >> -    ret = do_get_kexec_buffer(prop, len, &addr, &size);
> >> -    fdt_delprop(fdt, chosen_node, "linux,ima-kexec-buffer");
> >> -    if (ret)
> >> -            return;
> >> -
> >> -    ret = delete_fdt_mem_rsv(fdt, addr, size);
> >> -    if (!ret)
> >> -            pr_debug("Removed old IMA buffer reservation.\n");
> >> -}
> >> -
> >>   #ifdef CONFIG_IMA_KEXEC
> >>   /**
> >>    * arch_ima_add_kexec_buffer - do arch-specific steps to add the IMA buffer
> >>    *
> >> + * @image: kimage struct to set IMA buffer data
> >> + * @load_addr: Starting address where IMA buffer is loaded at
> >> + * @size: Number of bytes in the IMA buffer
> >> + *
> >>    * Architectures should use this function to pass on the IMA buffer
> >>    * information to the next kernel.
> >>    *
> >> @@ -179,7 +140,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)
> >>      int ret, addr_cells, size_cells, entry_size;
> >>      u8 value[16];
> >>
> >> -    remove_ima_buffer(fdt, chosen_node);
> >> +    remove_ima_kexec_buffer(fdt, chosen_node);
> >>      if (!image->arch.ima_buffer_size)
> >>              return 0;
> >>
> >> @@ -201,7 +162,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)
> >>      if (ret)
> >>              return ret;
> >>
> >> -    ret = fdt_setprop(fdt, chosen_node, "linux,ima-kexec-buffer", value,
> >> +    ret = fdt_setprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, value,
> >>                        entry_size);
> >>      if (ret < 0)
> >>              return -EINVAL;
> >> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c
> >> index 4602e467ca8b..d2e6f8ce0e42 100644
> >> --- a/drivers/of/fdt.c
> >> +++ b/drivers/of/fdt.c
> >> @@ -25,6 +25,7 @@
> >>   #include <linux/serial_core.h>
> >>   #include <linux/sysfs.h>
> >>   #include <linux/random.h>
> >> +#include <linux/kexec.h>
> >>
> >>   #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */
> >>   #include <asm/page.h>
> >> @@ -1289,4 +1290,113 @@ static int __init of_fdt_raw_init(void)
> >>   late_initcall(of_fdt_raw_init);
> >>   #endif
> >>
> >> +#ifdef CONFIG_HAVE_IMA_KEXEC
> >
> > Can we avoid #ifdef and use IS_ENABLED() within the functions?
> I can use IS_ENABLED() or move the functions to another C file and
> conditionally build based on the CONFIG.
>
> >
> >> +/**
> >> + * do_get_kexec_buffer - Get address and size of IMA kexec buffer
> >> + *
> >> + * @prop: IMA kexec buffer node in the device tree
> >> + * @len: Size of the given device tree node property
> >> + * @addr: Return address of the node
> >> + * @size: Return size of the node
> >> + */
> >> +int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
> >> +                    size_t *size)
> >> +{
> >> +    int addr_cells, size_cells;
> >> +    struct device_node *root;
> >> +
> >> +    root = of_find_node_by_path("/");
> >
> > The code in fdt.c operates on flat trees. This is an unflattened tree.
>
> Would it be better if I move these functions to a new C file under
> "drivers/of"?
>
> >
> >> +    if (!root)
> >> +            return -EINVAL;
> >> +
> >> +    addr_cells = of_n_addr_cells(root);
> >> +    size_cells = of_n_size_cells(root);
> >> +
> >> +    of_node_put(root);
> >> +
> >> +    if (len < 4 * (addr_cells + size_cells))
> >> +            return -ENOENT;
> >> +
> >> +    *addr = of_read_number(prop, addr_cells);
> >> +    *size = of_read_number(prop + 4 * addr_cells, size_cells);
> >
> > There's nothing in this function specific to 'kexec buffer'.
> >
> > This interface is kind of broken. 'prop' could come from anywhere in
> > the tree, but we always read the root address and size cells. Those only
> > apply to immediate child node properties. And anything other than
> > immediate root child nodes, there needs to be address translation.
> >
>
> I moved this function from arch/powerpc/kexec/ima.c to here so that it
> can be shared by ARM64 as well. Below is one of the usage of
> do_get_kexec_buffer().
>
>          prop = of_get_property(of_chosen, "linux,ima-kexec-buffer", &len);
>          if (!prop)
>                  return -ENOENT;
>
>          ret = do_get_kexec_buffer(prop, len, &tmp_addr, &tmp_size);
>
> Is there a way to detect if the 'prop' is an immediate root child node
> or not so that we can handle it appropriately in this interface function?

Just don't split up the function.

> >> +
> >> +    return 0;
> >> +}
> >> +
> >> +/**
> >> + * remove_ima_kexec_buffer - remove the IMA buffer property and
> >> + *                       reservation from @fdt
> >
> > IIRC, kerneldoc requires this to be one line.
> I'll upate.
>
> >
> >> + *
> >> + * @fdt: Flattened Device Tree to update
> >> + * @chosen_node: Offset to the chosen node in the device tree
> >> + *
> >> + * The IMA measurement buffer is of no use to a subsequent kernel,
> >> + * so we always remove it from the device tree.
> >> + */
> >> +void remove_ima_kexec_buffer(void *fdt, int chosen_node)
> >> +{
> >
> > Can't this go in some common kexec code?
>
> I'd considered moving this to "kernel", but Thiago suggested
> "drivers/of" is a better place. If you have suggestions for a better
> place, please let me know.

Humm, I guess drivers/of/ is fine, but it should be its own file.
Also, there's nothing arch specific about handling bootargs, initrd,
etc., so create a common function to handle all that. You can probably
include allocating and copying the new fdt as part of that. Powerpc
does kmalloc and arm64 does vmalloc for the new fdt. Is that a
necessary difference?

Then the IMA handling can all be private to the new file?

>
> Note that this function needs to be available even when CONFIG_IMA is
> not enabled. So I can't move it to "security/integrity/ima".
>
> >> +    int ret, len;
> >> +    unsigned long addr;
> >> +    size_t size;
> >> +    const void *prop;
> >> +
> >> +    prop = fdt_getprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, &len);
> >> +    if (!prop) {
> >> +            pr_debug("Unable to find the ima kexec buffer node\n");
> >> +            return;
> >> +    }
> >> +
> >> +    ret = do_get_kexec_buffer(prop, len, &addr, &size);
> >> +    fdt_delprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER);
> >> +    if (ret) {
> >> +            pr_err("Unable to delete the ima kexec buffer node\n");
> >> +            return;
> >> +    }
> >> +
> >> +    ret = delete_fdt_mem_rsv(fdt, addr, size);
> >> +    if (!ret)
> >> +            pr_debug("Removed old IMA buffer reservation.\n");
> >> +}
> >> +#endif /* CONFIG_HAVE_IMA_KEXEC */
> >> +
> >> +#ifdef CONFIG_KEXEC_FILE
> >> +/**
> >> + * delete_fdt_mem_rsv - delete memory reservation with given address and size
> >> + *
> >> + * @fdt: Flattened Device Tree to update
> >> + * @start: Starting address of the reservation to delete
> >> + * @size: Size of the reservation to delete
> >> + *
> >> + * Return: 0 on success, or negative errno on error.
> >> + */
> >> +int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)
> >> +{
> >> +    int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);
> >> +
> >> +    for (i = 0; i < num_rsvs; i++) {
> >> +            uint64_t rsv_start, rsv_size;
> >> +
> >> +            ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);
> >> +            if (ret) {
> >> +                    pr_err("Malformed device tree.\n");
> >> +                    return -EINVAL;
> >> +            }
> >> +
> >> +            if (rsv_start == start && rsv_size == size) {
> >> +                    ret = fdt_del_mem_rsv(fdt, i);
> >> +                    if (ret) {
> >> +                            pr_err("Error deleting device tree reservation.\n");
> >> +                            return -EINVAL;
> >> +                    }
> >> +
> >> +                    pr_debug("Freed reserved memory at %lu of size %lu\n",
> >> +                             start, size);
> >> +                    return 0;
> >> +            }
> >> +    }
> >> +
> >> +    return -ENOENT;
> >> +}
> >> +#endif /* CONFIG_KEXEC_FILE */
> >> +
> >>   #endif /* CONFIG_OF_EARLY_FLATTREE */
> >> diff --git a/include/linux/kexec.h b/include/linux/kexec.h
> >> index 9e93bef52968..6c6c6791a7ba 100644
> >> --- a/include/linux/kexec.h
> >> +++ b/include/linux/kexec.h
> >> @@ -407,6 +407,30 @@ static inline int kexec_crash_loaded(void) { return 0; }
> >>   #define kexec_in_progress false
> >>   #endif /* CONFIG_KEXEC_CORE */
> >>
> >> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_HAVE_IMA_KEXEC)
> >
> > CONFIG_OF_EARLY_FLATTREE is wrong because that's all early boot (i.e.
> > init section) functions.
> >
> > If these functions are implemented in fdt.c, then this is the wrong
> > header. But it's the implementation that should move.
>
> If you think defining these functions in "drivers/of" is okay, I'll move
> them to a new C file in OF.
>
> Would CONFIG_OF_FLATTREE be a better one to enable these ima kexec
> functions? "of_" and "fdt_" are the functions called by these ima kexec
> functions.

Yes.

>
> >
> >> +extern void remove_ima_kexec_buffer(void *fdt, int chosen_node);
> >> +extern int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,
> >> +                           size_t *size);
> >> +#else
> >> +static inline void remove_ima_kexec_buffer(void *fdt, int chosen_node) {}
> >> +static inline int do_get_kexec_buffer(const void *prop, int len,
> >> +                                  unsigned long *addr, size_t *size)
> >> +{
> >> +    return -EOPNOTSUPP;
> >> +}
> >> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_HAVE_IMA_KEXEC */
> >> +
> >> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_KEXEC_FILE)
> >> +extern int delete_fdt_mem_rsv(void *fdt, unsigned long start,
> >> +                          unsigned long size);
> >> +#else
> >> +static inline int delete_fdt_mem_rsv(void *fdt, unsigned long start,
> >> +                                 unsigned long size)
> >> +{
> >> +    return 0;
> >> +}
> >> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_KEXEC_FILE */
> >> +
> >>   #endif /* !defined(__ASSEBMLY__) */
> >>
> >>   #endif /* LINUX_KEXEC_H */
> >> diff --git a/include/linux/libfdt.h b/include/linux/libfdt.h
> >> index 90ed4ebfa692..75fb40aa013b 100644
> >> --- a/include/linux/libfdt.h
> >> +++ b/include/linux/libfdt.h
> >> @@ -5,4 +5,7 @@
> >>   #include <linux/libfdt_env.h>
> >>   #include "../../scripts/dtc/libfdt/libfdt.h"
> >>
> >> +/* Common device tree properties */
> >> +#define FDT_PROP_IMA_KEXEC_BUFFER   "linux,ima-kexec-buffer"
> >
> > This is not part of libfdt. We generally don't do defines for DT
> > strings.
> Should I create a new header file for this or would you suggest using an
> existing one?

I'm suggesting no define at all. The string is already descriptive, so
what do you gain with the define name? Nothing.

Rob
nramas Nov. 5, 2020, 9:54 p.m. UTC | #10
On 11/5/20 6:38 AM, Rob Herring wrote:

> On Wed, Nov 4, 2020 at 6:46 PM Lakshmi Ramasubramanian

> <nramas@linux.microsoft.com> wrote:

>>

>> On 11/4/20 2:28 PM, Rob Herring wrote:

>>

>> Hi Rob,

>>

>> Thanks for reviewing the patch.

>>

>>> On Fri, Oct 30, 2020 at 10:44:26AM -0700, Lakshmi Ramasubramanian wrote:

>>>> The functions remove_ima_buffer() and delete_fdt_mem_rsv() that handle

>>>> carrying forward the IMA measurement logs on kexec for powerpc do not

>>>> have architecture specific code, but they are currently defined for

>>>> powerpc only.

>>>>

>>>> remove_ima_buffer() and delete_fdt_mem_rsv() are used to remove

>>>> the IMA log entry from the device tree and free the memory reserved

>>>> for the log. These functions need to be defined even if the current

>>>> kernel does not support carrying forward IMA log across kexec since

>>>> the previous kernel could have supported that and therefore the current

>>>> kernel needs to free the allocation.

>>>>

>>>> Rename remove_ima_buffer() to remove_ima_kexec_buffer().

>>>> Define remove_ima_kexec_buffer() and delete_fdt_mem_rsv() in

>>>> drivers/of/fdt.c. A later patch in this series will use these functions

>>>> to free the allocation, if any, made by the previous kernel for ARM64.

>>>>

>>>> Define FDT_PROP_IMA_KEXEC_BUFFER for the chosen node, namely

>>>> "linux,ima-kexec-buffer", that is added to the DTB to hold

>>>> the address and the size of the memory reserved to carry

>>>> the IMA measurement log.

>>>>

>>>> Co-developed-by: Prakhar Srivastava <prsriva@linux.microsoft.com>

>>>> Signed-off-by: Prakhar Srivastava <prsriva@linux.microsoft.com>

>>>> Signed-off-by: Lakshmi Ramasubramanian <nramas@linux.microsoft.com>

>>>> Reported-by: kernel test robot <lkp@intel.com> error: kernel/kexec_file_fdt.c:30: undefined reference to `fdt_num_mem_rsv'

>>>

>>> This should be added for a commit fixing the reported problem. 0-day

>>> didn't report what this patch implements.

>> I'll remove that in the updated patch I'll post.

>>

>>>

>>>> ---

>>>>    arch/powerpc/include/asm/ima.h   |  10 +--

>>>>    arch/powerpc/include/asm/kexec.h |   1 -

>>>>    arch/powerpc/kexec/file_load.c   |  33 +---------

>>>>    arch/powerpc/kexec/ima.c         |  55 +++-------------

>>>>    drivers/of/fdt.c                 | 110 +++++++++++++++++++++++++++++++

>>>>    include/linux/kexec.h            |  24 +++++++

>>>>    include/linux/libfdt.h           |   3 +

>>>>    7 files changed, 149 insertions(+), 87 deletions(-)

>>>>

>>>> diff --git a/arch/powerpc/include/asm/ima.h b/arch/powerpc/include/asm/ima.h

>>>> index ead488cf3981..6355a85a3289 100644

>>>> --- a/arch/powerpc/include/asm/ima.h

>>>> +++ b/arch/powerpc/include/asm/ima.h

>>>> @@ -2,17 +2,13 @@

>>>>    #ifndef _ASM_POWERPC_IMA_H

>>>>    #define _ASM_POWERPC_IMA_H

>>>>

>>>> +#include <linux/kexec.h>

>>>> +

>>>>    struct kimage;

>>>>

>>>>    int ima_get_kexec_buffer(void **addr, size_t *size);

>>>>    int ima_free_kexec_buffer(void);

>>>>

>>>> -#ifdef CONFIG_IMA

>>>> -void remove_ima_buffer(void *fdt, int chosen_node);

>>>> -#else

>>>> -static inline void remove_ima_buffer(void *fdt, int chosen_node) {}

>>>> -#endif

>>>> -

>>>>    #ifdef CONFIG_IMA_KEXEC

>>>>    int arch_ima_add_kexec_buffer(struct kimage *image, unsigned long load_addr,

>>>>                             size_t size);

>>>> @@ -22,7 +18,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node);

>>>>    static inline int setup_ima_buffer(const struct kimage *image, void *fdt,

>>>>                                  int chosen_node)

>>>>    {

>>>> -    remove_ima_buffer(fdt, chosen_node);

>>>> +    remove_ima_kexec_buffer(fdt, chosen_node);

>>>>       return 0;

>>>>    }

>>>>    #endif /* CONFIG_IMA_KEXEC */

>>>> diff --git a/arch/powerpc/include/asm/kexec.h b/arch/powerpc/include/asm/kexec.h

>>>> index 55d6ede30c19..7c223031ecdd 100644

>>>> --- a/arch/powerpc/include/asm/kexec.h

>>>> +++ b/arch/powerpc/include/asm/kexec.h

>>>> @@ -126,7 +126,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,

>>>>    int setup_new_fdt(const struct kimage *image, void *fdt,

>>>>                 unsigned long initrd_load_addr, unsigned long initrd_len,

>>>>                 const char *cmdline);

>>>> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size);

>>>>

>>>>    #ifdef CONFIG_PPC64

>>>>    struct kexec_buf;

>>>> diff --git a/arch/powerpc/kexec/file_load.c b/arch/powerpc/kexec/file_load.c

>>>> index 9a232bc36c8f..7a17655c530e 100644

>>>> --- a/arch/powerpc/kexec/file_load.c

>>>> +++ b/arch/powerpc/kexec/file_load.c

>>>> @@ -18,6 +18,7 @@

>>>>    #include <linux/kexec.h>

>>>>    #include <linux/of_fdt.h>

>>>>    #include <linux/libfdt.h>

>>>> +#include <linux/kexec.h>

>>>>    #include <asm/setup.h>

>>>>    #include <asm/ima.h>

>>>>

>>>> @@ -109,38 +110,6 @@ int setup_purgatory(struct kimage *image, const void *slave_code,

>>>>       return 0;

>>>>    }

>>>>

>>>> -/**

>>>> - * delete_fdt_mem_rsv - delete memory reservation with given address and size

>>>> - *

>>>> - * Return: 0 on success, or negative errno on error.

>>>> - */

>>>> -int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)

>>>> -{

>>>> -    int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);

>>>> -

>>>> -    for (i = 0; i < num_rsvs; i++) {

>>>> -            uint64_t rsv_start, rsv_size;

>>>> -

>>>> -            ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);

>>>> -            if (ret) {

>>>> -                    pr_err("Malformed device tree.\n");

>>>> -                    return -EINVAL;

>>>> -            }

>>>> -

>>>> -            if (rsv_start == start && rsv_size == size) {

>>>> -                    ret = fdt_del_mem_rsv(fdt, i);

>>>> -                    if (ret) {

>>>> -                            pr_err("Error deleting device tree reservation.\n");

>>>> -                            return -EINVAL;

>>>> -                    }

>>>> -

>>>> -                    return 0;

>>>> -            }

>>>> -    }

>>>> -

>>>> -    return -ENOENT;

>>>> -}

>>>> -

>>>>    /*

>>>>     * setup_new_fdt - modify /chosen and memory reservation for the next kernel

>>>>     * @image:         kexec image being loaded.

>>>> diff --git a/arch/powerpc/kexec/ima.c b/arch/powerpc/kexec/ima.c

>>>> index 720e50e490b6..2b790230ea15 100644

>>>> --- a/arch/powerpc/kexec/ima.c

>>>> +++ b/arch/powerpc/kexec/ima.c

>>>> @@ -11,6 +11,8 @@

>>>>    #include <linux/of.h>

>>>>    #include <linux/memblock.h>

>>>>    #include <linux/libfdt.h>

>>>> +#include <linux/ima.h>

>>>> +#include <asm/ima.h>

>>>>

>>>>    static int get_addr_size_cells(int *addr_cells, int *size_cells)

>>>>    {

>>>> @@ -28,24 +30,6 @@ static int get_addr_size_cells(int *addr_cells, int *size_cells)

>>>>       return 0;

>>>>    }

>>>>

>>>> -static int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,

>>>> -                           size_t *size)

>>>> -{

>>>> -    int ret, addr_cells, size_cells;

>>>> -

>>>> -    ret = get_addr_size_cells(&addr_cells, &size_cells);

>>>> -    if (ret)

>>>> -            return ret;

>>>> -

>>>> -    if (len < 4 * (addr_cells + size_cells))

>>>> -            return -ENOENT;

>>>> -

>>>> -    *addr = of_read_number(prop, addr_cells);

>>>> -    *size = of_read_number(prop + 4 * addr_cells, size_cells);

>>>> -

>>>> -    return 0;

>>>> -}

>>>> -

>>>>    /**

>>>>     * ima_get_kexec_buffer - get IMA buffer from the previous kernel

>>>>     * @addr:  On successful return, set to point to the buffer contents.

>>>> @@ -100,37 +84,14 @@ int ima_free_kexec_buffer(void)

>>>>

>>>>    }

>>>>

>>>> -/**

>>>> - * remove_ima_buffer - remove the IMA buffer property and reservation from @fdt

>>>> - *

>>>> - * The IMA measurement buffer is of no use to a subsequent kernel, so we always

>>>> - * remove it from the device tree.

>>>> - */

>>>> -void remove_ima_buffer(void *fdt, int chosen_node)

>>>> -{

>>>> -    int ret, len;

>>>> -    unsigned long addr;

>>>> -    size_t size;

>>>> -    const void *prop;

>>>> -

>>>> -    prop = fdt_getprop(fdt, chosen_node, "linux,ima-kexec-buffer", &len);

>>>> -    if (!prop)

>>>> -            return;

>>>> -

>>>> -    ret = do_get_kexec_buffer(prop, len, &addr, &size);

>>>> -    fdt_delprop(fdt, chosen_node, "linux,ima-kexec-buffer");

>>>> -    if (ret)

>>>> -            return;

>>>> -

>>>> -    ret = delete_fdt_mem_rsv(fdt, addr, size);

>>>> -    if (!ret)

>>>> -            pr_debug("Removed old IMA buffer reservation.\n");

>>>> -}

>>>> -

>>>>    #ifdef CONFIG_IMA_KEXEC

>>>>    /**

>>>>     * arch_ima_add_kexec_buffer - do arch-specific steps to add the IMA buffer

>>>>     *

>>>> + * @image: kimage struct to set IMA buffer data

>>>> + * @load_addr: Starting address where IMA buffer is loaded at

>>>> + * @size: Number of bytes in the IMA buffer

>>>> + *

>>>>     * Architectures should use this function to pass on the IMA buffer

>>>>     * information to the next kernel.

>>>>     *

>>>> @@ -179,7 +140,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)

>>>>       int ret, addr_cells, size_cells, entry_size;

>>>>       u8 value[16];

>>>>

>>>> -    remove_ima_buffer(fdt, chosen_node);

>>>> +    remove_ima_kexec_buffer(fdt, chosen_node);

>>>>       if (!image->arch.ima_buffer_size)

>>>>               return 0;

>>>>

>>>> @@ -201,7 +162,7 @@ int setup_ima_buffer(const struct kimage *image, void *fdt, int chosen_node)

>>>>       if (ret)

>>>>               return ret;

>>>>

>>>> -    ret = fdt_setprop(fdt, chosen_node, "linux,ima-kexec-buffer", value,

>>>> +    ret = fdt_setprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, value,

>>>>                         entry_size);

>>>>       if (ret < 0)

>>>>               return -EINVAL;

>>>> diff --git a/drivers/of/fdt.c b/drivers/of/fdt.c

>>>> index 4602e467ca8b..d2e6f8ce0e42 100644

>>>> --- a/drivers/of/fdt.c

>>>> +++ b/drivers/of/fdt.c

>>>> @@ -25,6 +25,7 @@

>>>>    #include <linux/serial_core.h>

>>>>    #include <linux/sysfs.h>

>>>>    #include <linux/random.h>

>>>> +#include <linux/kexec.h>

>>>>

>>>>    #include <asm/setup.h>  /* for COMMAND_LINE_SIZE */

>>>>    #include <asm/page.h>

>>>> @@ -1289,4 +1290,113 @@ static int __init of_fdt_raw_init(void)

>>>>    late_initcall(of_fdt_raw_init);

>>>>    #endif

>>>>

>>>> +#ifdef CONFIG_HAVE_IMA_KEXEC

>>>

>>> Can we avoid #ifdef and use IS_ENABLED() within the functions?

>> I can use IS_ENABLED() or move the functions to another C file and

>> conditionally build based on the CONFIG.

>>

>>>

>>>> +/**

>>>> + * do_get_kexec_buffer - Get address and size of IMA kexec buffer

>>>> + *

>>>> + * @prop: IMA kexec buffer node in the device tree

>>>> + * @len: Size of the given device tree node property

>>>> + * @addr: Return address of the node

>>>> + * @size: Return size of the node

>>>> + */

>>>> +int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,

>>>> +                    size_t *size)

>>>> +{

>>>> +    int addr_cells, size_cells;

>>>> +    struct device_node *root;

>>>> +

>>>> +    root = of_find_node_by_path("/");

>>>

>>> The code in fdt.c operates on flat trees. This is an unflattened tree.

>>

>> Would it be better if I move these functions to a new C file under

>> "drivers/of"?

>>

>>>

>>>> +    if (!root)

>>>> +            return -EINVAL;

>>>> +

>>>> +    addr_cells = of_n_addr_cells(root);

>>>> +    size_cells = of_n_size_cells(root);

>>>> +

>>>> +    of_node_put(root);

>>>> +

>>>> +    if (len < 4 * (addr_cells + size_cells))

>>>> +            return -ENOENT;

>>>> +

>>>> +    *addr = of_read_number(prop, addr_cells);

>>>> +    *size = of_read_number(prop + 4 * addr_cells, size_cells);

>>>

>>> There's nothing in this function specific to 'kexec buffer'.

>>>

>>> This interface is kind of broken. 'prop' could come from anywhere in

>>> the tree, but we always read the root address and size cells. Those only

>>> apply to immediate child node properties. And anything other than

>>> immediate root child nodes, there needs to be address translation.

>>>

>>

>> I moved this function from arch/powerpc/kexec/ima.c to here so that it

>> can be shared by ARM64 as well. Below is one of the usage of

>> do_get_kexec_buffer().

>>

>>           prop = of_get_property(of_chosen, "linux,ima-kexec-buffer", &len);

>>           if (!prop)

>>                   return -ENOENT;

>>

>>           ret = do_get_kexec_buffer(prop, len, &tmp_addr, &tmp_size);

>>

>> Is there a way to detect if the 'prop' is an immediate root child node

>> or not so that we can handle it appropriately in this interface function?

> 

> Just don't split up the function.

ok

> 

>>>> +

>>>> +    return 0;

>>>> +}

>>>> +

>>>> +/**

>>>> + * remove_ima_kexec_buffer - remove the IMA buffer property and

>>>> + *                       reservation from @fdt

>>>

>>> IIRC, kerneldoc requires this to be one line.

>> I'll upate.

>>

>>>

>>>> + *

>>>> + * @fdt: Flattened Device Tree to update

>>>> + * @chosen_node: Offset to the chosen node in the device tree

>>>> + *

>>>> + * The IMA measurement buffer is of no use to a subsequent kernel,

>>>> + * so we always remove it from the device tree.

>>>> + */

>>>> +void remove_ima_kexec_buffer(void *fdt, int chosen_node)

>>>> +{

>>>

>>> Can't this go in some common kexec code?

>>

>> I'd considered moving this to "kernel", but Thiago suggested

>> "drivers/of" is a better place. If you have suggestions for a better

>> place, please let me know.

> 

> Humm, I guess drivers/of/ is fine, but it should be its own file.

> Also, there's nothing arch specific about handling bootargs, initrd,

> etc., so create a common function to handle all that. You can probably

> include allocating and copying the new fdt as part of that. Powerpc

> does kmalloc and arm64 does vmalloc for the new fdt. Is that a

> necessary difference?

> 


Agreed - I also think there's nothing arch specific about handling 
bootargs, initrd, etc. But refactoring the code that handles the above 
would be a much bigger change.

Would it be fine if we do it in two or more steps (to minimize 
regression risks)?

  #1 - move minimal functions/code to add support for ima buffer on 
kexec for ARM64 (addressed in this patch series)

  #2+ - refactor further to move more arch independent code to common 
functions in the next set of patche series.

> Then the IMA handling can all be private to the new file?

> 

>>

>> Note that this function needs to be available even when CONFIG_IMA is

>> not enabled. So I can't move it to "security/integrity/ima".

>>

>>>> +    int ret, len;

>>>> +    unsigned long addr;

>>>> +    size_t size;

>>>> +    const void *prop;

>>>> +

>>>> +    prop = fdt_getprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER, &len);

>>>> +    if (!prop) {

>>>> +            pr_debug("Unable to find the ima kexec buffer node\n");

>>>> +            return;

>>>> +    }

>>>> +

>>>> +    ret = do_get_kexec_buffer(prop, len, &addr, &size);

>>>> +    fdt_delprop(fdt, chosen_node, FDT_PROP_IMA_KEXEC_BUFFER);

>>>> +    if (ret) {

>>>> +            pr_err("Unable to delete the ima kexec buffer node\n");

>>>> +            return;

>>>> +    }

>>>> +

>>>> +    ret = delete_fdt_mem_rsv(fdt, addr, size);

>>>> +    if (!ret)

>>>> +            pr_debug("Removed old IMA buffer reservation.\n");

>>>> +}

>>>> +#endif /* CONFIG_HAVE_IMA_KEXEC */

>>>> +

>>>> +#ifdef CONFIG_KEXEC_FILE

>>>> +/**

>>>> + * delete_fdt_mem_rsv - delete memory reservation with given address and size

>>>> + *

>>>> + * @fdt: Flattened Device Tree to update

>>>> + * @start: Starting address of the reservation to delete

>>>> + * @size: Size of the reservation to delete

>>>> + *

>>>> + * Return: 0 on success, or negative errno on error.

>>>> + */

>>>> +int delete_fdt_mem_rsv(void *fdt, unsigned long start, unsigned long size)

>>>> +{

>>>> +    int i, ret, num_rsvs = fdt_num_mem_rsv(fdt);

>>>> +

>>>> +    for (i = 0; i < num_rsvs; i++) {

>>>> +            uint64_t rsv_start, rsv_size;

>>>> +

>>>> +            ret = fdt_get_mem_rsv(fdt, i, &rsv_start, &rsv_size);

>>>> +            if (ret) {

>>>> +                    pr_err("Malformed device tree.\n");

>>>> +                    return -EINVAL;

>>>> +            }

>>>> +

>>>> +            if (rsv_start == start && rsv_size == size) {

>>>> +                    ret = fdt_del_mem_rsv(fdt, i);

>>>> +                    if (ret) {

>>>> +                            pr_err("Error deleting device tree reservation.\n");

>>>> +                            return -EINVAL;

>>>> +                    }

>>>> +

>>>> +                    pr_debug("Freed reserved memory at %lu of size %lu\n",

>>>> +                             start, size);

>>>> +                    return 0;

>>>> +            }

>>>> +    }

>>>> +

>>>> +    return -ENOENT;

>>>> +}

>>>> +#endif /* CONFIG_KEXEC_FILE */

>>>> +

>>>>    #endif /* CONFIG_OF_EARLY_FLATTREE */

>>>> diff --git a/include/linux/kexec.h b/include/linux/kexec.h

>>>> index 9e93bef52968..6c6c6791a7ba 100644

>>>> --- a/include/linux/kexec.h

>>>> +++ b/include/linux/kexec.h

>>>> @@ -407,6 +407,30 @@ static inline int kexec_crash_loaded(void) { return 0; }

>>>>    #define kexec_in_progress false

>>>>    #endif /* CONFIG_KEXEC_CORE */

>>>>

>>>> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_HAVE_IMA_KEXEC)

>>>

>>> CONFIG_OF_EARLY_FLATTREE is wrong because that's all early boot (i.e.

>>> init section) functions.

>>>

>>> If these functions are implemented in fdt.c, then this is the wrong

>>> header. But it's the implementation that should move.

>>

>> If you think defining these functions in "drivers/of" is okay, I'll move

>> them to a new C file in OF.

>>

>> Would CONFIG_OF_FLATTREE be a better one to enable these ima kexec

>> functions? "of_" and "fdt_" are the functions called by these ima kexec

>> functions.

> 

> Yes.

Thanks - I will use CONFIG_OF_FLATTREE to handle this change.

> 

>>

>>>

>>>> +extern void remove_ima_kexec_buffer(void *fdt, int chosen_node);

>>>> +extern int do_get_kexec_buffer(const void *prop, int len, unsigned long *addr,

>>>> +                           size_t *size);

>>>> +#else

>>>> +static inline void remove_ima_kexec_buffer(void *fdt, int chosen_node) {}

>>>> +static inline int do_get_kexec_buffer(const void *prop, int len,

>>>> +                                  unsigned long *addr, size_t *size)

>>>> +{

>>>> +    return -EOPNOTSUPP;

>>>> +}

>>>> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_HAVE_IMA_KEXEC */

>>>> +

>>>> +#if defined(CONFIG_OF_EARLY_FLATTREE) && defined(CONFIG_KEXEC_FILE)

>>>> +extern int delete_fdt_mem_rsv(void *fdt, unsigned long start,

>>>> +                          unsigned long size);

>>>> +#else

>>>> +static inline int delete_fdt_mem_rsv(void *fdt, unsigned long start,

>>>> +                                 unsigned long size)

>>>> +{

>>>> +    return 0;

>>>> +}

>>>> +#endif /* CONFIG_OF_EARLY_FLATTREE && CONFIG_KEXEC_FILE */

>>>> +

>>>>    #endif /* !defined(__ASSEBMLY__) */

>>>>

>>>>    #endif /* LINUX_KEXEC_H */

>>>> diff --git a/include/linux/libfdt.h b/include/linux/libfdt.h

>>>> index 90ed4ebfa692..75fb40aa013b 100644

>>>> --- a/include/linux/libfdt.h

>>>> +++ b/include/linux/libfdt.h

>>>> @@ -5,4 +5,7 @@

>>>>    #include <linux/libfdt_env.h>

>>>>    #include "../../scripts/dtc/libfdt/libfdt.h"

>>>>

>>>> +/* Common device tree properties */

>>>> +#define FDT_PROP_IMA_KEXEC_BUFFER   "linux,ima-kexec-buffer"

>>>

>>> This is not part of libfdt. We generally don't do defines for DT

>>> strings.

>> Should I create a new header file for this or would you suggest using an

>> existing one?

> 

> I'm suggesting no define at all. The string is already descriptive, so

> what do you gain with the define name? Nothing.

> 


Sure - I will remove the #define and use "linux,ima-kexec-buffer" directly.

  -lakshmi