Hi, 在 2024/03/19 10:13, Matthew Sakai 写道: > > On 3/18/24 21:43, Yu Kuai wrote: >> Hi, >> >> 在 2024/03/19 9:18, Yu Kuai 写道: >>> Hi, >>> >>> 在 2024/03/18 17:39, Christian Brauner 写道: >>>> On Sat, Mar 16, 2024 at 10:49:33AM +0800, Yu Kuai wrote: >>>>> Hi, Christian >>>>> >>>>> 在 2024/03/15 21:54, Christian Brauner 写道: >>>>>> On Fri, Mar 15, 2024 at 08:08:49PM +0800, Yu Kuai wrote: >>>>>>> Hi, Christian >>>>>>> Hi, Christoph >>>>>>> Hi, Jan >>>>>>> >>>>>>> Perhaps now is a good time to send a formal version of this set. >>>>>>> However, I'm not sure yet what branch should I rebase and send >>>>>>> this set. >>>>>>> Should I send to the vfs tree? >>>>>> >>>>>> Nearly all of it is in fs/ so I'd say yes. >>>>>> . >>>>> >>>>> I see that you just create a new branch vfs.fixes, perhaps can I >>>>> rebase >>>>> this set against this branch? >>>> >>>> Please base it on vfs.super. I'll rebase it to v6.9-rc1 on Sunday. >>> >>> Okay, I just see that vfs.super doesn't contain commit >>> 1cdeac6da33f("btrfs: pass btrfs_device to btrfs_scratch_superblocks()"), >>> and you might need to fix the conflict at some point. >> >> And there is another problem, dm-vdo doesn't exist in vfs.super yet. Do >> you still want me to rebase here? >> > > The dm-vdo changes don't appear to rely on earlier patches in the > series, so I think dm-vdo could incorporate the dm-vdo patch > independently from the rest of the series, if that would be helpful. (I > don't want to confuse things too much.) In that case it would go through > the dm tree with the rest of dm-vdo. We want to remove the 'bd_inode' field in this set. And if we want to go through dm tree for dm-vdo changes, we must keep the field for now. I don't have preference, Christian will make the decision. 😉 Thanks, Kuai > > Matt > > . >
On 3/18/24 21:43, Yu Kuai wrote:
> Hi,
>
> 在 2024/03/19 9:18, Yu Kuai 写道:
>> Hi,
>>
>> 在 2024/03/18 17:39, Christian Brauner 写道:
>>> On Sat, Mar 16, 2024 at 10:49:33AM +0800, Yu Kuai wrote:
>>>> Hi, Christian
>>>>
>>>> 在 2024/03/15 21:54, Christian Brauner 写道:
>>>>> On Fri, Mar 15, 2024 at 08:08:49PM +0800, Yu Kuai wrote:
>>>>>> Hi, Christian
>>>>>> Hi, Christoph
>>>>>> Hi, Jan
>>>>>>
>>>>>> Perhaps now is a good time to send a formal version of this set.
>>>>>> However, I'm not sure yet what branch should I rebase and send
>>>>>> this set.
>>>>>> Should I send to the vfs tree?
>>>>>
>>>>> Nearly all of it is in fs/ so I'd say yes.
>>>>> .
>>>>
>>>> I see that you just create a new branch vfs.fixes, perhaps can I rebase
>>>> this set against this branch?
>>>
>>> Please base it on vfs.super. I'll rebase it to v6.9-rc1 on Sunday.
>>
>> Okay, I just see that vfs.super doesn't contain commit
>> 1cdeac6da33f("btrfs: pass btrfs_device to btrfs_scratch_superblocks()"),
>> and you might need to fix the conflict at some point.
>
> And there is another problem, dm-vdo doesn't exist in vfs.super yet. Do
> you still want me to rebase here?
>
The dm-vdo changes don't appear to rely on earlier patches in the
series, so I think dm-vdo could incorporate the dm-vdo patch
independently from the rest of the series, if that would be helpful. (I
don't want to confuse things too much.) In that case it would go through
the dm tree with the rest of dm-vdo.
Matt
On 3/18/24 16:54, Guenter Roeck wrote:
> On 3/18/24 13:37, Kenneth Raeburn wrote:
>> (resend because of accidental HTML lossage)
>>
>> On Thu, Mar 14, 2024 at 7:38 PM Guenter Roeck <linux@roeck-us.net> wrote:
>>> On sparc64, with gcc 11.4, the above code results in:
>>>
>>> ERROR: modpost: "__bswapdi2" [drivers/md/dm-vdo/dm-vdo.ko] undefined!
>>>
>>> Guenter
>>
>> Thanks for catching that. I don't think our team has any sparc
>> machines readily available for testing.
>> This is an artifact of our having imported user-mode code to use in
>> the kernel. We should probably be using le64_to_cpup and friends, as
>> we do elsewhere, so it doesn't try to pull in libgcc support routines.
>>
>
> I am kind of getting wary about reporting such issues. Should I drop
> building dm-vdo images for sparc ? Would it be possible to add
> "depends on BROKEN if SPARC" configuration option to indicate that
> the code isn't expected to be buildable on sparc, much less work ?
First off, sorry for the slow response. I was away for the last couple
of days.
I would encourage you to report issues when you find them.
Theoretically, VDO should work on sparc. However since we don't have
that architecture readily available to us, we won't know what to fix
unless it gets reported.
In this case, we really shouldn't be relying on these libgcc builtins
anyway. I hope to have a patch in a couple of days that you can try out.
If it turns out there are more extensive issues with VDO on sparc, I am
open to disabling the config, but I hope it won't come to that.
Matt
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 4.19-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-4.19 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit c048b053cf419f832c332a213fe4c3438cf2f192 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 569904f73994a..95ed46930a90e 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -46,11 +46,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 6e65ec0e627a6..04ef89e318564 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -72,11 +72,11 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned n_blocks; - struct bvec_iter iter; - struct work_struct work; /*
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 5.4-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-5.4 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit 51de750d029a1fe516b67d3bc7c0fb63a9962b47 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 6b4fb4e7506f8..6866aa2aade3b 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -46,11 +46,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 74ad36b6dbf53..fee7c7a81ce4e 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -73,11 +73,11 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned n_blocks; - struct bvec_iter iter; - struct work_struct work; /*
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 5.10-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-5.10 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit 0c7560792b0b22d4ba1f8d36963820feb9bfceb5 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 5edcdcee91c23..5deda6c6fa2e7 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -48,11 +48,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 78d1e51195ada..f61c89c79cf5b 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -74,11 +74,11 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned n_blocks; - struct bvec_iter iter; - struct work_struct work; /*
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 5.15-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-5.15 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit 7089a433a187c2234ed43f0001780d009b151036 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 5904af05f6057..a3db76dcdb3ec 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -49,11 +49,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 78d1e51195ada..f61c89c79cf5b 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -74,11 +74,11 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned n_blocks; - struct bvec_iter iter; - struct work_struct work; /*
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 6.1-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-6.1 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit 320d818da76028188f19af4d0232c4bc2f4785f5 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 3e215aa85b99a..e8c534b5870ac 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -52,11 +52,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 4620a98c99561..db93a91169d5e 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -80,12 +80,12 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned int n_blocks; bool in_tasklet; - struct bvec_iter iter; - struct work_struct work; char *recheck_buffer;
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 6.6-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-6.6 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit 833fa53b7154fc172b264066368ad7c527b225a0 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 5a296db23feb3..aa6bb5b4704ba 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -53,11 +53,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 4620a98c99561..db93a91169d5e 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -80,12 +80,12 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned int n_blocks; bool in_tasklet; - struct bvec_iter iter; - struct work_struct work; char *recheck_buffer;
This is a note to let you know that I've just added the patch titled dm-verity, dm-crypt: align "struct bvec_iter" correctly to the 6.7-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: dm-verity-dm-crypt-align-struct-bvec_iter-correctly.patch and it can be found in the queue-6.7 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@vger.kernel.org> know about it. commit cd786f8debe865b168e35b2ce0ae1723e3b04ba9 Author: Mikulas Patocka <mpatocka@redhat.com> Date: Tue Feb 20 19:11:51 2024 +0100 dm-verity, dm-crypt: align "struct bvec_iter" correctly [ Upstream commit 787f1b2800464aa277236a66eb3c279535edd460 ] "struct bvec_iter" is defined with the __packed attribute, so it is aligned on a single byte. On X86 (and on other architectures that support unaligned addresses in hardware), "struct bvec_iter" is accessed using the 8-byte and 4-byte memory instructions, however these instructions are less efficient if they operate on unaligned addresses. (on RISC machines that don't have unaligned access in hardware, GCC generates byte-by-byte accesses that are very inefficient - see [1]) This commit reorders the entries in "struct dm_verity_io" and "struct convert_context", so that "struct bvec_iter" is aligned on 8 bytes. [1] https://lore.kernel.org/all/ZcLuWUNRZadJr0tQ@fedora/T/ Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Signed-off-by: Mike Snitzer <snitzer@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> diff --git a/drivers/md/dm-crypt.c b/drivers/md/dm-crypt.c index 4ab4e8dcfd3e2..35f50193959e0 100644 --- a/drivers/md/dm-crypt.c +++ b/drivers/md/dm-crypt.c @@ -53,11 +53,11 @@ struct convert_context { struct completion restart; struct bio *bio_in; - struct bio *bio_out; struct bvec_iter iter_in; + struct bio *bio_out; struct bvec_iter iter_out; - u64 cc_sector; atomic_t cc_pending; + u64 cc_sector; union { struct skcipher_request *req; struct aead_request *req_aead; diff --git a/drivers/md/dm-verity.h b/drivers/md/dm-verity.h index 4620a98c99561..db93a91169d5e 100644 --- a/drivers/md/dm-verity.h +++ b/drivers/md/dm-verity.h @@ -80,12 +80,12 @@ struct dm_verity_io { /* original value of bio->bi_end_io */ bio_end_io_t *orig_bi_end_io; + struct bvec_iter iter; + sector_t block; unsigned int n_blocks; bool in_tasklet; - struct bvec_iter iter; - struct work_struct work; char *recheck_buffer;
On 3/18/2024 1:08 AM, Roberto Sassu wrote:
> On Sun, 2024-03-17 at 22:17 -0700, Eric Biggers wrote:
>> On Fri, Mar 15, 2024 at 08:35:48PM -0700, Fan Wu wrote:
>>> +config IPE_PROP_FS_VERITY
>>> + bool "Enable property for fs-verity files"
>>> + depends on FS_VERITY && FS_VERITY_BUILTIN_SIGNATURES
>>> + help
>>> + This option enables the usage of properties "fsverity_signature"
>>> + and "fsverity_digest". These properties evaluate to TRUE when
>>> + a file is fsverity enabled and with a signed digest
>>
>> Again: why would anyone care if there is a signature, if that signature is not
>> checked.
>>
>> I think you meant to write something like: "when a file is fsverity enabled and
>> has a valid builtin signature whose signing cert is in the .fs-verity keyring".
>
> I was also thinking the same. I didn't follow the recent development
> closely, but unless IPE locks somehow the .fs-verity keyring, the
> property you suggested would not be immutable. Meaning that someone can
> add/remove a key in that keyring, making the property true or false.
>
> Roberto
Yes, the .fs-verity keyring's mutability could affect the property's
immutability. However, we are not planing to "lock" the keyrings, but we
would like to use policies languages to express what certificate can be
trusted.
For example, we can have a rule like this:
#Certificate declaration
CERTIFICATE=MyCertificate CertThumbprint=DummyThumbprint
op=EXECUTE fsverity_signature=MyCertificate action=ALLOW
This will be our immediate next work after the initial version is accepted.
-Fan
On 3/18/24 13:37, Kenneth Raeburn wrote:
> (resend because of accidental HTML lossage)
>
> On Thu, Mar 14, 2024 at 7:38 PM Guenter Roeck <linux@roeck-us.net> wrote:
>> On sparc64, with gcc 11.4, the above code results in:
>>
>> ERROR: modpost: "__bswapdi2" [drivers/md/dm-vdo/dm-vdo.ko] undefined!
>>
>> Guenter
>
> Thanks for catching that. I don't think our team has any sparc
> machines readily available for testing.
> This is an artifact of our having imported user-mode code to use in
> the kernel. We should probably be using le64_to_cpup and friends, as
> we do elsewhere, so it doesn't try to pull in libgcc support routines.
>
I am kind of getting wary about reporting such issues. Should I drop
building dm-vdo images for sparc ? Would it be possible to add
"depends on BROKEN if SPARC" configuration option to indicate that
the code isn't expected to be buildable on sparc, much less work ?
Thanks,
Guenter
Hello: This series was applied to jaegeuk/f2fs.git (dev) by Jens Axboe <axboe@kernel.dk>: On Sun, 28 Jan 2024 23:52:15 -0800 you wrote: > Fueled by the LSFMM discussion on removing GFP_NOFS initiated by Willy, > I've looked into the sole GFP_NOFS allocation in zonefs. As it turned out, > it is only done for zone management commands and can be removed. > > After digging into more callers of blkdev_zone_mgmt() I came to the > conclusion that the gfp_mask parameter can be removed alltogether. > > [...] Here is the summary with links: - [f2fs-dev,v3,1/5] zonefs: pass GFP_KERNEL to blkdev_zone_mgmt() call https://git.kernel.org/jaegeuk/f2fs/c/9105ce591b42 - [f2fs-dev,v3,2/5] dm: dm-zoned: guard blkdev_zone_mgmt with noio scope https://git.kernel.org/jaegeuk/f2fs/c/218082010ace - [f2fs-dev,v3,3/5] btrfs: zoned: call blkdev_zone_mgmt in nofs scope https://git.kernel.org/jaegeuk/f2fs/c/d9d556755f16 - [f2fs-dev,v3,4/5] f2fs: guard blkdev_zone_mgmt with nofs scope https://git.kernel.org/jaegeuk/f2fs/c/147ec1c60e32 - [f2fs-dev,v3,5/5] block: remove gfp_flags from blkdev_zone_mgmt https://git.kernel.org/jaegeuk/f2fs/c/71f4ecdbb42a You are awesome, thank you! -- Deet-doot-dot, I am a bot. https://korg.docs.kernel.org/patchwork/pwbot.html
On 3/17/2024 10:17 PM, Eric Biggers wrote:
> On Fri, Mar 15, 2024 at 08:35:48PM -0700, Fan Wu wrote:
>> +config IPE_PROP_FS_VERITY
>> + bool "Enable property for fs-verity files"
>> + depends on FS_VERITY && FS_VERITY_BUILTIN_SIGNATURES
>> + help
>> + This option enables the usage of properties "fsverity_signature"
>> + and "fsverity_digest". These properties evaluate to TRUE when
>> + a file is fsverity enabled and with a signed digest
>
> Again: why would anyone care if there is a signature, if that signature is not
> checked.
>
> I think you meant to write something like: "when a file is fsverity enabled and
> has a valid builtin signature whose signing cert is in the .fs-verity keyring".
>
> - Eric
Thanks for the suggestion. I agree this is a more accurate description.
I'll update the description to include these details.
-Fan
(resend because of accidental HTML lossage)
On Thu, Mar 14, 2024 at 7:38 PM Guenter Roeck <linux@roeck-us.net> wrote:
> On sparc64, with gcc 11.4, the above code results in:
>
> ERROR: modpost: "__bswapdi2" [drivers/md/dm-vdo/dm-vdo.ko] undefined!
>
> Guenter
Thanks for catching that. I don't think our team has any sparc
machines readily available for testing.
This is an artifact of our having imported user-mode code to use in
the kernel. We should probably be using le64_to_cpup and friends, as
we do elsewhere, so it doesn't try to pull in libgcc support routines.
Ken
When we recheck the data after checksum failure, and the recheck is successful, we would leak the "checksums" pointer. Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Fixes: c88f5e553fe3 ("dm-integrity: recheck the integrity tag after a failure") Cc: stable@vger.kernel.org --- drivers/md/dm-integrity.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) Index: linux-2.6/drivers/md/dm-integrity.c =================================================================== --- linux-2.6.orig/drivers/md/dm-integrity.c 2024-03-11 12:53:59.000000000 +0100 +++ linux-2.6/drivers/md/dm-integrity.c 2024-03-18 17:40:57.000000000 +0100 @@ -1848,12 +1848,12 @@ again: r = dm_integrity_rw_tag(ic, checksums, &dio->metadata_block, &dio->metadata_offset, checksums_ptr - checksums, dio->op == REQ_OP_READ ? TAG_CMP : TAG_WRITE); if (unlikely(r)) { + if (likely(checksums != checksums_onstack)) + kfree(checksums); if (r > 0) { - integrity_recheck(dio, checksums); + integrity_recheck(dio, checksums_onstack); goto skip_io; } - if (likely(checksums != checksums_onstack)) - kfree(checksums); goto error; }
On Sun, 2024-03-17 at 22:17 -0700, Eric Biggers wrote:
> On Fri, Mar 15, 2024 at 08:35:48PM -0700, Fan Wu wrote:
> > +config IPE_PROP_FS_VERITY
> > + bool "Enable property for fs-verity files"
> > + depends on FS_VERITY && FS_VERITY_BUILTIN_SIGNATURES
> > + help
> > + This option enables the usage of properties "fsverity_signature"
> > + and "fsverity_digest". These properties evaluate to TRUE when
> > + a file is fsverity enabled and with a signed digest
>
> Again: why would anyone care if there is a signature, if that signature is not
> checked.
>
> I think you meant to write something like: "when a file is fsverity enabled and
> has a valid builtin signature whose signing cert is in the .fs-verity keyring".
I was also thinking the same. I didn't follow the recent development
closely, but unless IPE locks somehow the .fs-verity keyring, the
property you suggested would not be immutable. Meaning that someone can
add/remove a key in that keyring, making the property true or false.
Roberto
On Fri, Mar 15, 2024 at 08:35:47PM -0700, Fan Wu wrote: > fsverity represents a mechanism to support both integrity and > authenticity protection of a file, supporting both signed and unsigned > digests. > > An LSM which controls access to a resource based on authenticity and > integrity of said resource, can then use this data to make an informed > decision on the authorization (provided by the LSM's policy) of said > claim. > > This effectively allows the extension of a policy enforcement layer in > LSM for fsverity, allowing for more granular control of how a > particular authenticity claim can be used. For example, "all (built-in) > signed fsverity files should be allowed to execute, but only these > hashes are allowed to be loaded as kernel modules". > > This enforcement must be done in kernel space, as a userspace only > solution would fail a simple litmus test: Download a self-contained > malicious binary that never touches the userspace stack. This > binary would still be able to execute. > > This patch introduces a new security_inode_setintegrity() hook call in > fsverity to store the verified fsverity signature in the inode's LSM blobs. > The hook call is executed after fsverity_verify_signature(), ensuring that > the signature is verified against fsverity's keyring. > > The last commit in this patch set will add a link to the IPE documentation in > fsverity.rst. There are multiple types of fsverity signatures. Please make it clear which one you're talking about ("fsverity builtin signatures"). > diff --git a/Documentation/filesystems/fsverity.rst b/Documentation/filesystems/fsverity.rst > index 13e4b18e5dbb..708c79631f32 100644 > --- a/Documentation/filesystems/fsverity.rst > +++ b/Documentation/filesystems/fsverity.rst > @@ -461,7 +461,10 @@ Enabling this option adds the following: > > 3. A new sysctl "fs.verity.require_signatures" is made available. > When set to 1, the kernel requires that all verity files have a > - correctly signed digest as described in (2). > + correctly signed digest as described in (2). Note that verification > + happens as long as the file's signature exists regardless of the state of > + "fs.verity.require_signatures", and the IPE LSM relies on this behavior > + to save verified signature into LSM blobs. This probably should go in item (2) instead, as that already mentions the behavior when opening a file. > The data that the signature as described in (2) must be a signature of > is the fs-verity file digest in the following format:: > @@ -481,7 +484,7 @@ be carefully considered before using them: > > - Builtin signature verification does *not* make the kernel enforce > that any files actually have fs-verity enabled. Thus, it is not a > - complete authentication policy. Currently, if it is used, the only > + complete authentication policy. Currently, if it is used, one > way to complete the authentication policy is for trusted userspace > code to explicitly check whether files have fs-verity enabled with a > signature before they are accessed. (With > @@ -489,6 +492,11 @@ be carefully considered before using them: > enabled suffices.) But, in this case the trusted userspace code > could just store the signature alongside the file and verify it > itself using a cryptographic library, instead of using this feature. > + Another approach is to utilize built-in signature verification in > + conjunction with the IPE LSM, which supports defining > + a kernel-enforced, system-wide authentication policy that allows only > + files with an fs-verity signature enabled to perform certain operations, > + such as execution. Note that IPE doesn't require fs.verity.require_signatures=1. Make this a new paragraph. Also, the acronym "IPE" should be spelled out somewhere. IPE should be mentioned in the list in the "Use cases" section next to IMA, and it can be spelled out there. > +#ifdef CONFIG_FS_VERITY_BUILTIN_SIGNATURES > +static int fsverity_inode_setintegrity(struct inode *inode, > + const struct fsverity_descriptor *desc) > +{ > + return security_inode_setintegrity(inode, LSM_INTGR_FSV_SIG, > + desc->signature, > + le32_to_cpu(desc->sig_size)); > +} > +#else > +static inline int fsverity_inode_setintegrity(struct inode *inode, > + const struct fsverity_descriptor *desc) > +{ > + return 0; > +} > +#endif /* CONFIG_FS_VERITY_BUILTIN_SIGNATURES */ > + > /* > * Create a new fsverity_info from the given fsverity_descriptor (with optional > * appended builtin signature), and check the signature if present. The > * fsverity_descriptor must have already undergone basic validation. > */ > -struct fsverity_info *fsverity_create_info(const struct inode *inode, > +struct fsverity_info *fsverity_create_info(struct inode *inode, > struct fsverity_descriptor *desc) > { > struct fsverity_info *vi; > @@ -241,6 +258,11 @@ struct fsverity_info *fsverity_create_info(const struct inode *inode, > } > } > > + err = fsverity_inode_setintegrity(inode, desc); > + > + if (err) > + goto fail; Delete the blank line before 'if (err)' > diff --git a/fs/verity/signature.c b/fs/verity/signature.c > index 90c07573dd77..d4ed03a114e9 100644 > --- a/fs/verity/signature.c > +++ b/fs/verity/signature.c > @@ -41,7 +41,10 @@ static struct key *fsverity_keyring; > * @sig_size: size of signature in bytes, or 0 if no signature > * > * If the file includes a signature of its fs-verity file digest, verify it > - * against the certificates in the fs-verity keyring. > + * against the certificates in the fs-verity keyring. Note that verification > + * happens as long as the file's signature exists regardless of the state of > + * fsverity_require_signatures, and the IPE LSM relies on this behavior > + * to save the verified file signature of the file into security blobs. "save the verified file signature of the file into security blobs" isn't what IPE actually does, though. And even if it was, it would not explain why IPE expects the signature to be verified. > diff --git a/include/linux/security.h b/include/linux/security.h > index 0885866b261e..edd12c0a673a 100644 > --- a/include/linux/security.h > +++ b/include/linux/security.h > @@ -86,6 +86,7 @@ enum lsm_event { > enum lsm_intgr_type { > LSM_INTGR_DMV_SIG, > LSM_INTGR_DMV_ROOTHASH, > + LSM_INTGR_FSV_SIG, > __LSM_INTGR_MAX > }; These are hard to understand because they are abbreviated too much. And again, there are multiple type of fsverity signatures. How about: enum lsm_integrity_type { LSM_INTEGRITY_DM_VERITY_SIG, LSM_INTEGRITY_DM_VERITY_ROOT_HASH, LSM_INTEGRITY_FS_VERITY_BUILTIN_SIG, __LSM_INTEGRITY_MAX }; - Eric
On Fri, Mar 15, 2024 at 08:35:48PM -0700, Fan Wu wrote:
> +config IPE_PROP_FS_VERITY
> + bool "Enable property for fs-verity files"
> + depends on FS_VERITY && FS_VERITY_BUILTIN_SIGNATURES
> + help
> + This option enables the usage of properties "fsverity_signature"
> + and "fsverity_digest". These properties evaluate to TRUE when
> + a file is fsverity enabled and with a signed digest
Again: why would anyone care if there is a signature, if that signature is not
checked.
I think you meant to write something like: "when a file is fsverity enabled and
has a valid builtin signature whose signing cert is in the .fs-verity keyring".
- Eric
On Sun, Mar 17, 2024 at 8:29 PM Casey Schaufler <casey@schaufler-ca.com> wrote:
> On 3/15/2024 8:35 PM, Fan Wu wrote:
> > This patch introduces a new hook to notify security system that the
> > content of initramfs has been unpacked into the rootfs.
> >
> > Upon receiving this notification, the security system can activate
> > a policy to allow only files that originated from the initramfs to
> > execute or load into kernel during the early stages of booting.
> >
> > This approach is crucial for minimizing the attack surface by
> > ensuring that only trusted files from the initramfs are operational
> > in the critical boot phase.
> >
> > Signed-off-by: Fan Wu <wufan@linux.microsoft.com>
> >
> > ---
> > v1-v11:
> > + Not present
> >
> > v12:
> > + Introduced
> >
> > v13:
> > + Rename the hook name to initramfs_populated()
> >
> > v14:
> > + No changes
> >
> > v15:
> > + No changes
> > ---
> > include/linux/lsm_hook_defs.h | 2 ++
> > include/linux/security.h | 8 ++++++++
> > init/initramfs.c | 3 +++
> > security/security.c | 10 ++++++++++
> > 4 files changed, 23 insertions(+)
> >
> > diff --git a/include/linux/lsm_hook_defs.h b/include/linux/lsm_hook_defs.h
> > index 334e00efbde4..7db99ae75651 100644
> > --- a/include/linux/lsm_hook_defs.h
> > +++ b/include/linux/lsm_hook_defs.h
> > @@ -450,3 +450,5 @@ LSM_HOOK(int, 0, uring_override_creds, const struct cred *new)
> > LSM_HOOK(int, 0, uring_sqpoll, void)
> > LSM_HOOK(int, 0, uring_cmd, struct io_uring_cmd *ioucmd)
> > #endif /* CONFIG_IO_URING */
> > +
> > +LSM_HOOK(void, LSM_RET_VOID, initramfs_populated, void)
>
> This is an awfully expensive way to set a flag. Adding a LSM hook list
> isn't free. Isn't there a way to capture this state change through one of
> the mount hooks?
Unfortunately no, the initramfs isn't mounted like a traditional
filesystem, it is "populated" by unpacking the cpio into the initramfs
at early boot. This LSM hook should be called exactly once during
boot, and the performance impact should be minimal; I should also be
wildly more performant than earlier revisions of this patchset that
required grabbing a single spinlock on every file access.
Of course if you have an idea on how this could be done
differently/better I think we're all open to new ideas ...
--
paul-moore.com
On 3/15/2024 8:35 PM, Fan Wu wrote: > This patch introduces a new hook to notify security system that the > content of initramfs has been unpacked into the rootfs. > > Upon receiving this notification, the security system can activate > a policy to allow only files that originated from the initramfs to > execute or load into kernel during the early stages of booting. > > This approach is crucial for minimizing the attack surface by > ensuring that only trusted files from the initramfs are operational > in the critical boot phase. > > Signed-off-by: Fan Wu <wufan@linux.microsoft.com> > > --- > v1-v11: > + Not present > > v12: > + Introduced > > v13: > + Rename the hook name to initramfs_populated() > > v14: > + No changes > > v15: > + No changes > --- > include/linux/lsm_hook_defs.h | 2 ++ > include/linux/security.h | 8 ++++++++ > init/initramfs.c | 3 +++ > security/security.c | 10 ++++++++++ > 4 files changed, 23 insertions(+) > > diff --git a/include/linux/lsm_hook_defs.h b/include/linux/lsm_hook_defs.h > index 334e00efbde4..7db99ae75651 100644 > --- a/include/linux/lsm_hook_defs.h > +++ b/include/linux/lsm_hook_defs.h > @@ -450,3 +450,5 @@ LSM_HOOK(int, 0, uring_override_creds, const struct cred *new) > LSM_HOOK(int, 0, uring_sqpoll, void) > LSM_HOOK(int, 0, uring_cmd, struct io_uring_cmd *ioucmd) > #endif /* CONFIG_IO_URING */ > + > +LSM_HOOK(void, LSM_RET_VOID, initramfs_populated, void) This is an awfully expensive way to set a flag. Adding a LSM hook list isn't free. Isn't there a way to capture this state change through one of the mount hooks? > diff --git a/include/linux/security.h b/include/linux/security.h > index 41a8f667bdfa..14fff542f2e3 100644 > --- a/include/linux/security.h > +++ b/include/linux/security.h > @@ -2255,4 +2255,12 @@ static inline int security_uring_cmd(struct io_uring_cmd *ioucmd) > #endif /* CONFIG_SECURITY */ > #endif /* CONFIG_IO_URING */ > > +#ifdef CONFIG_SECURITY > +extern void security_initramfs_populated(void); > +#else > +static inline void security_initramfs_populated(void) > +{ > +} > +#endif /* CONFIG_SECURITY */ > + > #endif /* ! __LINUX_SECURITY_H */ > diff --git a/init/initramfs.c b/init/initramfs.c > index da79760b8be3..cc9115117896 100644 > --- a/init/initramfs.c > +++ b/init/initramfs.c > @@ -17,6 +17,7 @@ > #include <linux/namei.h> > #include <linux/init_syscalls.h> > #include <linux/umh.h> > +#include <linux/security.h> > > #include "do_mounts.h" > > @@ -719,6 +720,8 @@ static void __init do_populate_rootfs(void *unused, async_cookie_t cookie) > #endif > } > > + security_initramfs_populated(); > + > done: > /* > * If the initrd region is overlapped with crashkernel reserved region, > diff --git a/security/security.c b/security/security.c > index 287bfac6b471..b10230c51c0b 100644 > --- a/security/security.c > +++ b/security/security.c > @@ -5675,3 +5675,13 @@ int security_uring_cmd(struct io_uring_cmd *ioucmd) > return call_int_hook(uring_cmd, ioucmd); > } > #endif /* CONFIG_IO_URING */ > + > +/** > + * security_initramfs_populated() - Notify LSMs that initramfs has been loaded > + * > + * Tells the LSMs the initramfs has been unpacked into the rootfs. > + */ > +void security_initramfs_populated(void) > +{ > + call_void_hook(initramfs_populated); > +}
From: Deven Bowers <deven.desai@linux.microsoft.com> Add IPE's admin and developer documentation to the kernel tree. Co-developed-by: Fan Wu <wufan@linux.microsoft.com> Signed-off-by: Deven Bowers <deven.desai@linux.microsoft.com> Signed-off-by: Fan Wu <wufan@linux.microsoft.com> --- v2: + No Changes v3: + Add Acked-by + Fixup code block syntax + Fix a minor grammatical issue. v4: + Update documentation with the results of other code changes. v5: + No changes v6: + No changes v7: + Add additional developer-level documentation + Update admin-guide docs to reflect changes. + Drop Acked-by due to significant changes + Added section about audit events in admin-guide v8: + Correct terminology from "audit event" to "audit record" + Add associated documentation with the correct "audit event" terminology. + Add some context to the historical motivation for IPE and design philosophy. + Add some content about the securityfs layout in the policies directory. + Various spelling and grammatical corrections. v9: + Correct spelling of "pitfalls" + Update the docs w.r.t the new parser and new audit formats v10: + Refine user docs per upstream suggestions + Update audit events part v11: + No changes v12: + Update audit formats + Update initramfs related docs + Add test suite link v13: + No changes v14: + No changes v15: + Update boot_verified part + Fix format issues + Add IPE doc link to fsverity.rst --- Documentation/admin-guide/LSM/index.rst | 1 + Documentation/admin-guide/LSM/ipe.rst | 774 ++++++++++++++++++ .../admin-guide/kernel-parameters.txt | 12 + Documentation/filesystems/fsverity.rst | 1 + Documentation/security/index.rst | 1 + Documentation/security/ipe.rst | 444 ++++++++++ MAINTAINERS | 2 + 7 files changed, 1235 insertions(+) create mode 100644 Documentation/admin-guide/LSM/ipe.rst create mode 100644 Documentation/security/ipe.rst diff --git a/Documentation/admin-guide/LSM/index.rst b/Documentation/admin-guide/LSM/index.rst index a6ba95fbaa9f..ce63be6d64ad 100644 --- a/Documentation/admin-guide/LSM/index.rst +++ b/Documentation/admin-guide/LSM/index.rst @@ -47,3 +47,4 @@ subdirectories. tomoyo Yama SafeSetID + ipe diff --git a/Documentation/admin-guide/LSM/ipe.rst b/Documentation/admin-guide/LSM/ipe.rst new file mode 100644 index 000000000000..27496ce32f16 --- /dev/null +++ b/Documentation/admin-guide/LSM/ipe.rst @@ -0,0 +1,774 @@ +.. SPDX-License-Identifier: GPL-2.0 + +Integrity Policy Enforcement (IPE) +================================== + +.. NOTE:: + + This is the documentation for admins, system builders, or individuals + attempting to use IPE. If you're looking for more developer-focused + documentation about IPE please see Documentation/security/ipe.rst + +Overview +-------- + +IPE is a Linux Security Module which takes a complimentary approach to +access control. Whereas existing mandatory access control mechanisms +base their decisions on labels and paths, IPE instead determines +whether or not an operation should be allowed based on immutable +security properties of the system component the operation is being +performed on. + +IPE itself does not mandate how the security property should be +evaluated, but relies on an extensible set of external property providers +to evaluate the component. IPE makes its decision based on reference +values for the selected properties, specified in the IPE policy. + +The reference values represent the value that the policy writer and the +local system administrator (based on the policy signature) trust for the +system to accomplish the desired tasks. + +One such provider is for example dm-verity, which is able to represent +the integrity property of a partition (its immutable state) with a digest. + +To enable IPE, ensure that ``CONFIG_SECURITY_IPE`` (under +:menuselection:`Security -> Integrity Policy Enforcement (IPE)`) config +option is enabled. + +Use Cases +--------- + +IPE works best in fixed-function devices: devices in which their purpose +is clearly defined and not supposed to be changed (e.g. network firewall +device in a data center, an IoT device, etcetera), where all software and +configuration is built and provisioned by the system owner. + +IPE is a long-way off for use in general-purpose computing: the Linux +community as a whole tends to follow a decentralized trust model (known as +the web of trust), which IPE has no support for it yet. Instead, IPE +supports PKI (public key infrastructure), which generally designates a +set of trusted entities that provide a measure of absolute trust. + +Additionally, while most packages are signed today, the files inside +the packages (for instance, the executables), tend to be unsigned. This +makes it difficult to utilize IPE in systems where a package manager is +expected to be functional, without major changes to the package manager +and ecosystem behind it. + +DIGLIM [#diglim]_ is a system that when combined with IPE, could be used to +enable and support general-purpose computing use cases. + +Known Limitations +----------------- + +IPE cannot verify the integrity of anonymous executable memory, such as +the trampolines created by gcc closures and libffi (<3.4.2), or JIT'd code. +Unfortunately, as this is dynamically generated code, there is no way +for IPE to ensure the integrity of this code to form a trust basis. In all +cases, the return result for these operations will be whatever the admin +configures as the ``DEFAULT`` action for ``EXECUTE``. + +IPE cannot verify the integrity of programs written in interpreted +languages when these scripts are invoked by passing these program files +to the interpreter. This is because the way interpreters execute these +files; the scripts themselves are not evaluated as executable code +through one of IPE's hooks, but they are merely text files that are read +(as opposed to compiled executables) [#interpreters]_. + +Threat Model +------------ + +The threat type addressed by IPE is tampering of executable userspace +code beyond the initially booted kernel, and the initial verification of +kernel modules that are loaded in userspace through ``modprobe`` or +``insmod``. + +A bare-minimum example of a threat that should be mitigated by IPE, is +an untrusted (potentially malicious) binary that is downloaded and +bundled with all required dependencies (including a loader, libc, etc). +With IPE, this binary should not be allowed to be executed, not even any +of its dependencies. + +Tampering violates integrity, yet lack of trust is caused by being +unable to detect tampering (and by extent verifying the integrity). +IPE's role in mitigating this threat is to verify the integrity (and +authenticity) of all executable code and to deny their use if they +cannot be trusted (as integrity verification fails, or the authorization +check fails against the reference value in the policy). IPE generates +audit logs which may be utilized to detect and analyze failures +resulting from policy violation. + +Tampering threat scenarios include modification or replacement of +executable code by a range of actors including: + +- Actors with physical access to the hardware +- Actors with local network access to the system +- Actors with access to the deployment system +- Compromised internal systems under external control +- Malicious end users of the system +- Compromised end users of the system +- Remote (external) compromise of the system + +IPE does not mitigate threats arising from malicious but authorized +developers (with access to a signing certificate), or compromised +developer tools used by them (i.e. return-oriented programming attacks). +Additionally, IPE draws hard security boundary between userspace and +kernelspace. As a result, IPE does not provide any protections against a +kernel level exploit, and a kernel-level exploit can disable or tamper +with IPE's protections. + +Policy +------ + +IPE policy is a plain-text [#devdoc]_ policy composed of multiple statements +over several lines. There is one required line, at the top of the +policy, indicating the policy name, and the policy version, for +instance:: + + policy_name=Ex_Policy policy_version=0.0.0 + +The policy name is a unique key identifying this policy in a human +readable name. This is used to create nodes under securityfs as well as +uniquely identify policies to deploy new policies vs update existing +policies. + +The policy version indicates the current version of the policy (NOT the +policy syntax version). This is used to prevent rollback of policy to +potentially insecure previous versions of the policy. + +The next portion of IPE policy are rules. Rules are formed by key=value +pairs, known as properties. IPE rules require two properties: ``action``, +which determines what IPE does when it encounters a match against the +rule, and ``op``, which determines when the rule should be evaluated. +The ordering is significant, a rule must start with ``op``, and end with +``action``. Thus, a minimal rule is:: + + op=EXECUTE action=ALLOW + +This example will allow any execution. Additional properties are used to +restrict attributes about the files being evaluated. These properties +are intended to be descriptions of systems within the kernel that can +provide a measure of integrity verification, such that IPE can determine +the trust of the resource based on the value of the property. + +Rules are evaluated top-to-bottom. As a result, any revocation rules, +or denies should be placed early in the file to ensure that these rules +are evaluated before a rule with ``action=ALLOW``. + +IPE policy supports comments. The character '#' will function as a +comment, ignoring all characters to the right of '#' until the newline. + +The default behavior of IPE evaluations can also be expressed in policy, +through the ``DEFAULT`` statement. This can be done at a global level, +or a per-operation level:: + + # Global + DEFAULT action=ALLOW + + # Operation Specific + DEFAULT op=EXECUTE action=ALLOW + +A default must be set for all known operations in IPE. If you want to +preserve older policies being compatible with newer kernels that can introduce +new operations, set a global default of ``ALLOW``, then override the +defaults on a per-operation basis (as above). + +With configurable policy-based LSMs, there's several issues with +enforcing the configurable policies at startup, around reading and +parsing the policy: + +1. The kernel *should* not read files from userspace, so directly reading + the policy file is prohibited. +2. The kernel command line has a character limit, and one kernel module + should not reserve the entire character limit for its own + configuration. +3. There are various boot loaders in the kernel ecosystem, so handing + off a memory block would be costly to maintain. + +As a result, IPE has addressed this problem through a concept of a "boot +policy". A boot policy is a minimal policy which is compiled into the +kernel. This policy is intended to get the system to a state where +userspace is set up and ready to receive commands, at which point a more +complex policy can be deployed via securityfs. The boot policy can be +specified via ``SECURITY_IPE_BOOT_POLICY`` config option, which accepts +a path to a plain-text version of the IPE policy to apply. This policy +will be compiled into the kernel. If not specified, IPE will be disabled +until a policy is deployed and activated through securityfs. + +Deploying Policies +~~~~~~~~~~~~~~~~~~ + +Policies can be deployed from userspace through securityfs. These policies +are signed through the PKCS#7 message format to enforce some level of +authorization of the policies (prohibiting an attacker from gaining +unconstrained root, and deploying an "allow all" policy). These +policies must be signed by a certificate that chains to the +``SYSTEM_TRUSTED_KEYRING``. With openssl, the policy can be signed by:: + + openssl smime -sign \ + -in "$MY_POLICY" \ + -signer "$MY_CERTIFICATE" \ + -inkey "$MY_PRIVATE_KEY" \ + -noattr \ + -nodetach \ + -nosmimecap \ + -outform der \ + -out "$MY_POLICY.p7b" + +Deploying the policies is done through securityfs, through the +``new_policy`` node. To deploy a policy, simply cat the file into the +securityfs node:: + + cat "$MY_POLICY.p7b" > /sys/kernel/security/ipe/new_policy + +Upon success, this will create one subdirectory under +``/sys/kernel/security/ipe/policies/``. The subdirectory will be the +``policy_name`` field of the policy deployed, so for the example above, +the directory will be ``/sys/kernel/security/ipe/policies/Ex_Policy``. +Within this directory, there will be seven files: ``pkcs7``, ``policy``, +``name``, ``version``, ``active``, ``update``, and ``delete``. + +The ``pkcs7`` file is read-only. Reading it returns the raw PKCS#7 data +that was provided to the kernel, representing the policy. If the policy being +read is the boot policy, this will return ``ENOENT``, as it is not signed. + +The ``policy`` file is read only. Reading it returns the PKCS#7 inner +content of the policy, which will be the plain text policy. + +The ``active`` file is used to set a policy as the currently active policy. +This file is rw, and accepts a value of ``"1"`` to set the policy as active. +Since only a single policy can be active at one time, all other policies +will be marked inactive. The policy being marked active must have a policy +version greater or equal to the currently-running version. + +The ``update`` file is used to update a policy that is already present +in the kernel. This file is write-only and accepts a PKCS#7 signed +policy. Two checks will always be performed on this policy: First, the +``policy_names`` must match with the updated version and the existing +version. Second the updated policy must have a policy version greater than +or equal to the currently-running version. This is to prevent rollback attacks. + +The ``delete`` file is used to remove a policy that is no longer needed. +This file is write-only and accepts a value of ``1`` to delete the policy. +On deletion, the securityfs node representing the policy will be removed. +However, delete the current active policy is not allowed and will return +an operation not permitted error. + +Similarly, writing to both ``update`` and ``new_policy`` could result in +bad message(policy syntax error) or file exists error. The latter error happens +when trying to deploy a policy with a ``policy_name`` while the kernel already +has a deployed policy with the same ``policy_name``. + +Deploying a policy will *not* cause IPE to start enforcing the policy. IPE will +only enforce the policy marked active. Note that only one policy can be active +at a time. + +Once deployment is successful, the policy can be activated, by writing file +``/sys/kernel/security/ipe/$policy_name/active``. +For example, the ``Ex_Policy`` can be activated by:: + + echo 1 > "/sys/kernel/security/ipe/Ex_Policy/active" + +From above point on, ``Ex_Policy`` is now the enforced policy on the +system. + +IPE also provides a way to delete policies. This can be done via the +``delete`` securityfs node, ``/sys/kernel/security/ipe/$policy_name/delete``. +Writing ``1`` to that file deletes the policy:: + + echo 1 > "/sys/kernel/security/ipe/$policy_name/delete" + +There is only one requirement to delete a policy: the policy being deleted +must be inactive. + +.. NOTE:: + + If a traditional MAC system is enabled (SELinux, apparmor, smack), all + writes to ipe's securityfs nodes require ``CAP_MAC_ADMIN``. + +Modes +~~~~~ + +IPE supports two modes of operation: permissive (similar to SELinux's +permissive mode) and enforced. In permissive mode, all events are +checked and policy violations are logged, but the policy is not really +enforced. This allows users to test policies before enforcing them. + +The default mode is enforce, and can be changed via the kernel command +line parameter ``ipe.enforce=(0|1)``, or the securityfs node +``/sys/kernel/security/ipe/enforce``. + +.. NOTE:: + + If a traditional MAC system is enabled (SELinux, apparmor, smack, etcetera), + all writes to ipe's securityfs nodes require ``CAP_MAC_ADMIN``. + +Audit Events +~~~~~~~~~~~~ + +1420 AUDIT_IPE_ACCESS +^^^^^^^^^^^^^^^^^^^^^ +Event Examples:: + + type=1420 audit(1653364370.067:61): ipe_op=EXECUTE ipe_hook=MMAP enforcing=1 pid=2241 comm="ld-linux.so" path="/deny/lib/libc.so.6" dev="sda2" ino=14549020 rule="DEFAULT action=DENY" + type=1300 audit(1653364370.067:61): SYSCALL arch=c000003e syscall=9 success=no exit=-13 a0=7f1105a28000 a1=195000 a2=5 a3=812 items=0 ppid=2219 pid=2241 auid=0 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts0 ses=2 comm="ld-linux.so" exe="/tmp/ipe-test/lib/ld-linux.so" subj=unconfined key=(null) + type=1327 audit(1653364370.067:61): 707974686F6E3300746573742F6D61696E2E7079002D6E00 + + type=1420 audit(1653364735.161:64): ipe_op=EXECUTE ipe_hook=MMAP enforcing=1 pid=2472 comm="mmap_test" path=? dev=? ino=? rule="DEFAULT action=DENY" + type=1300 audit(1653364735.161:64): SYSCALL arch=c000003e syscall=9 success=no exit=-13 a0=0 a1=1000 a2=4 a3=21 items=0 ppid=2219 pid=2472 auid=0 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts0 ses=2 comm="mmap_test" exe="/root/overlake_test/upstream_test/vol_fsverity/bin/mmap_test" subj=unconfined key=(null) + type=1327 audit(1653364735.161:64): 707974686F6E3300746573742F6D61696E2E7079002D6E00 + +This event indicates that IPE made an access control decision; the IPE +specific record (1420) is always emitted in conjunction with a +``AUDITSYSCALL`` record. + +Determining whether IPE is in permissive or enforced mode can be derived +from ``success`` property and exit code of the ``AUDITSYSCALL`` record. + + +Field descriptions: + ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| Field | Value Type | Optional? | Description of Value | ++===========+============+===========+=================================================================================+ +| ipe_op | string | No | The IPE operation name associated with the log | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| ipe_hook | string | No | The name of the LSM hook that triggered the IPE event | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| enforcing | integer | No | The current IPE enforcing state 1 is in enforcing mode, 0 is in permissive mode | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| pid | integer | No | The pid of the process that triggered the IPE event. | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| comm | string | No | The command line program name of the process that triggered the IPE event | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| path | string | Yes | The absolute path to the evaluated file | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| ino | integer | Yes | The inode number of the evaluated file | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| dev | string | Yes | The device name of the evaluated file, e.g. vda | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ +| rule | string | No | The matched policy rule | ++-----------+------------+-----------+---------------------------------------------------------------------------------+ + +1421 AUDIT_IPE_CONFIG_CHANGE +^^^^^^^^^^^^^^^^^^^^^^^^^^^^ + +Event Example:: + + type=1421 audit(1653425583.136:54): old_active_pol_name="Allow_All" old_active_pol_version=0.0.0 old_policy_digest=sha256:E3B0C44298FC1C149AFBF4C8996FB92427AE41E4649B934CA495991B7852B855 new_active_pol_name="boot_verified" new_active_pol_version=0.0.0 new_policy_digest=sha256:820EEA5B40CA42B51F68962354BA083122A20BB846F26765076DD8EED7B8F4DB auid=4294967295 ses=4294967295 lsm=ipe res=1 + type=1300 audit(1653425583.136:54): SYSCALL arch=c000003e syscall=1 success=yes exit=2 a0=3 a1=5596fcae1fb0 a2=2 a3=2 items=0 ppid=184 pid=229 auid=4294967295 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts0 ses=4294967295 comm="python3" exe="/usr/bin/python3.10" key=(null) + type=1327 audit(1653425583.136:54): PROCTITLE proctitle=707974686F6E3300746573742F6D61696E2E7079002D66002E2 + +This event indicates that IPE switched the active poliy from one to another +along with the version and the hash digest of the two policies. +Note IPE can only have one policy active at a time, all access decision +evaluation is based on the current active policy. +The normal procedure to deploy a new policy is loading the policy to deploy +into the kernel first, then switch the active policy to it. + +This record will always be emitted in conjunction with a ``AUDITSYSCALL`` record for the ``write`` syscall. + +Field descriptions: + ++------------------------+------------+-----------+---------------------------------------------------+ +| Field | Value Type | Optional? | Description of Value | ++========================+============+===========+===================================================+ +| old_active_pol_name | string | No | The name of previous active policy | ++------------------------+------------+-----------+---------------------------------------------------+ +| old_active_pol_version | string | No | The version of previous active policy | ++------------------------+------------+-----------+---------------------------------------------------+ +| old_policy_digest | string | No | The hash of previous active policy | ++------------------------+------------+-----------+---------------------------------------------------+ +| new_active_pol_name | string | No | The name of current active policy | ++------------------------+------------+-----------+---------------------------------------------------+ +| new_active_pol_version | string | No | The version of current active policy | ++------------------------+------------+-----------+---------------------------------------------------+ +| new_policy_digest | string | No | The hash of current active policy | ++------------------------+------------+-----------+---------------------------------------------------+ +| auid | integer | No | The login user ID | ++------------------------+------------+-----------+---------------------------------------------------+ +| ses | integer | No | The login session ID | ++------------------------+------------+-----------+---------------------------------------------------+ +| lsm | string | No | The lsm name associated with the event | ++------------------------+------------+-----------+---------------------------------------------------+ +| res | integer | No | The result of the audited operation(success/fail) | ++------------------------+------------+-----------+---------------------------------------------------+ + +1422 AUDIT_IPE_POLICY_LOAD +^^^^^^^^^^^^^^^^^^^^^^^^^^ + +Event Example:: + + type=1422 audit(1653425529.927:53): policy_name="boot_verified" policy_version=0.0.0 policy_digest=sha256:820EEA5B40CA42B51F68962354BA083122A20BB846F26765076DD8EED7B8F4DB auid=4294967295 ses=4294967295 lsm=ipe res=1 + type=1300 audit(1653425529.927:53): arch=c000003e syscall=1 success=yes exit=2567 a0=3 a1=5596fcae1fb0 a2=a07 a3=2 items=0 ppid=184 pid=229 auid=4294967295 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=pts0 ses=4294967295 comm="python3" exe="/usr/bin/python3.10" key=(null) + type=1327 audit(1653425529.927:53): PROCTITLE proctitle=707974686F6E3300746573742F6D61696E2E7079002D66002E2E + +This record indicates a new policy has been loaded into the kernel with the policy name, policy version and policy hash. + +This record will always be emitted in conjunction with a ``AUDITSYSCALL`` record for the ``write`` syscall. + +Field descriptions: + ++----------------+------------+-----------+---------------------------------------------------+ +| Field | Value Type | Optional? | Description of Value | ++================+============+===========+===================================================+ +| policy_name | string | No | The policy_name | ++----------------+------------+-----------+---------------------------------------------------+ +| policy_version | string | No | The policy_version | ++----------------+------------+-----------+---------------------------------------------------+ +| policy_digest | string | No | The policy hash | ++----------------+------------+-----------+---------------------------------------------------+ +| auid | integer | No | The login user ID | ++----------------+------------+-----------+---------------------------------------------------+ +| ses | integer | No | The login session ID | ++----------------+------------+-----------+---------------------------------------------------+ +| lsm | string | No | The lsm name associated with the event | ++----------------+------------+-----------+---------------------------------------------------+ +| res | integer | No | The result of the audited operation(success/fail) | ++----------------+------------+-----------+---------------------------------------------------+ + + +1404 AUDIT_MAC_STATUS +^^^^^^^^^^^^^^^^^^^^^ + +Event Examples:: + + type=1404 audit(1653425689.008:55): enforcing=0 old_enforcing=1 auid=4294967295 ses=4294967295 enabled=1 old-enabled=1 lsm=ipe res=1 + type=1300 audit(1653425689.008:55): arch=c000003e syscall=1 success=yes exit=2 a0=1 a1=55c1065e5c60 a2=2 a3=0 items=0 ppid=405 pid=441 auid=0 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=) + type=1327 audit(1653425689.008:55): proctitle="-bash" + + type=1404 audit(1653425689.008:55): enforcing=1 old_enforcing=0 auid=4294967295 ses=4294967295 enabled=1 old-enabled=1 lsm=ipe res=1 + type=1300 audit(1653425689.008:55): arch=c000003e syscall=1 success=yes exit=2 a0=1 a1=55c1065e5c60 a2=2 a3=0 items=0 ppid=405 pid=441 auid=0 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=) + type=1327 audit(1653425689.008:55): proctitle="-bash" + +This record will always be emitted in conjunction with a ``AUDITSYSCALL`` record for the ``write`` syscall. + +Field descriptions: + ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| Field | Value Type | Optional? | Description of Value | ++===============+============+===========+=================================================================================================+ +| enforcing | integer | No | The enforcing state IPE is being switched to, 1 is in enforcing mode, 0 is in permissive mode | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| old_enforcing | integer | No | The enforcing state IPE is being switched from, 1 is in enforcing mode, 0 is in permissive mode | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| auid | integer | No | The login user ID | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| ses | integer | No | The login session ID | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| enabled | integer | No | The new TTY audit enabled setting | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| old-enabled | integer | No | The old TTY audit enabled setting | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| lsm | string | No | The lsm name associated with the event | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ +| res | integer | No | The result of the audited operation(success/fail) | ++---------------+------------+-----------+-------------------------------------------------------------------------------------------------+ + + +Success Auditing +^^^^^^^^^^^^^^^^ + +IPE supports success auditing. When enabled, all events that pass IPE +policy and are not blocked will emit an audit event. This is disabled by +default, and can be enabled via the kernel command line +``ipe.success_audit=(0|1)`` or +``/sys/kernel/security/ipe/success_audit`` securityfs file. + +This is *very* noisy, as IPE will check every userspace binary on the +system, but is useful for debugging policies. + +.. NOTE:: + + If a traditional MAC system is enabled (SELinux, apparmor, smack, etcetera), + all writes to ipe's securityfs nodes require ``CAP_MAC_ADMIN``. + +Properties +---------- + +As explained above, IPE properties are ``key=value`` pairs expressed in +IPE policy. Two properties are built-into the policy parser: 'op' and +'action'. The other properties are deterministic attributes to express +across files. Currently those properties are: '``boot_verified``', +'``dmverity_signature``', '``dmverity_roothash``', '``fsverity_signature``', +'``fsverity_digest``'. A description of all properties supported by IPE +are listed below: + +op +~~ + +Indicates the operation for a rule to apply to. Must be in every rule, +as the first token. IPE supports the following operations: + + ``EXECUTE`` + + Pertains to any file attempting to be executed, or loaded as an + executable. + + ``FIRMWARE``: + + Pertains to firmware being loaded via the firmware_class interface. + This covers both the preallocated buffer and the firmware file + itself. + + ``KMODULE``: + + Pertains to loading kernel modules via ``modprobe`` or ``insmod``. + + ``KEXEC_IMAGE``: + + Pertains to kernel images loading via ``kexec``. + + ``KEXEC_INITRAMFS`` + + Pertains to initrd images loading via ``kexec --initrd``. + + ``POLICY``: + + Controls loading policies via reading a kernel-space initiated read. + + An example of such is loading IMA policies by writing the path + to the policy file to ``$securityfs/ima/policy`` + + ``X509_CERT``: + + Controls loading IMA certificates through the Kconfigs, + ``CONFIG_IMA_X509_PATH`` and ``CONFIG_EVM_X509_PATH``. + +action +~~~~~~ + + Determines what IPE should do when a rule matches. Must be in every + rule, as the final clause. Can be one of: + + ``ALLOW``: + + If the rule matches, explicitly allow access to the resource to proceed + without executing any more rules. + + ``DENY``: + + If the rule matches, explicitly prohibit access to the resource to + proceed without executing any more rules. + +boot_verified +~~~~~~~~~~~~~ + + This property can be utilized for authorization of files from initramfs. + The format of this property is:: + + boot_verified=(TRUE|FALSE) + + + .. WARNING:: + + This property will trust files from initramfs(rootfs). It should + only be used during early booting stage. Before mounting the real + rootfs on top of the initramfs, initramfs + script will recursively remove all files and directories on the + initramfs. This is typically implemented by using switch_root(8) [#switch_root]_. + Therefore the initramfs will be empty and not accessible after the real + rootfs takes over. It is advised to switch to a different policy + that doesn't rely on the property after this point. + This ensures that the trust policies remain relevant and effective + throughout the system's operation. + +dmverity_roothash +~~~~~~~~~~~~~~~~~ + + This property can be utilized for authorization or revocation of + specific dm-verity volumes, identified via sroot hash. It has a + dependency on the DM_VERITY module. This property is controlled by + the ``IPE_PROP_DM_VERITY`` config option, it will be automatically + selected when ``IPE_SECURITY`` , ``DM_VERITY`` and + ``DM_VERITY_VERIFY_ROOTHASH_SIG`` are all enabled. + The format of this property is:: + + dmverity_roothash=DigestName:HexadecimalString + + The supported DigestNames for dmverity_roothash are [#dmveritydigests]_ [#securedigest]_ : + + + blake2b-512 + + blake2s-256 + + sha1 + + sha256 + + sha384 + + sha512 + + sha3-224 + + sha3-256 + + sha3-384 + + sha3-512 + + md4 + + md5 + + sm3 + + rmd160 + +dmverity_signature +~~~~~~~~~~~~~~~~~~ + + This property can be utilized for authorization of all dm-verity + volumes that have a signed roothash that chains to a keyring + specified by dm-verity's configuration, either the system trusted + keyring, or the secondary keyring. It depends on + ``DM_VERITY_VERIFY_ROOTHASH_SIG`` config option and is controlled by + the ``IPE_PROP_DM_VERITY`` config option, it will be automatically + selected when ``IPE_SECURITY``, ``DM_VERITY`` and + ``DM_VERITY_VERIFY_ROOTHASH_SIG`` are all enabled. + The format of this property is:: + + dmverity_signature=(TRUE|FALSE) + +fsverity_digest +~~~~~~~~~~~~~~~ + + This property can be utilized for authorization or revocation of + specific fsverity enabled file, identified via its fsverity digest. + It depends on ``FS_VERITY`` config option and is controlled by + ``CONFIG_IPE_PROP_FS_VERITY``. The format of this property is:: + + fsverity_digest=DigestName:HexadecimalString + + The supported DigestNames for fsverity_roothash are [#fsveritydigest]_ [#securedigest]_ : + + + sha256 + + sha512 + +fsverity_signature +~~~~~~~~~~~~~~~~~~ + + This property can be utilized for authorization of all fsverity + enabled files that is verified by fsverity. The keyring that the + signature is verified against is subject to fsverity's configuration, + typically the fsverity keyring. It depends on + ``CONFIG_FS_VERITY_BUILTIN_SIGNATURES`` and it is controlled by + the Kconfig ``CONFIG_IPE_PROP_FS_VERITY``. The format of this + property is:: + + fsverity_signature=(TRUE|FALSE) + +Policy Examples +--------------- + +Allow all +~~~~~~~~~ + +:: + + policy_name=Allow_All policy_version=0.0.0 + DEFAULT action=ALLOW + +Allow only initramfs +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + policy_name=Allow_All_Initramfs policy_version=0.0.0 + DEFAULT action=DENY + + op=EXECUTE boot_verified=TRUE action=ALLOW + +Allow any signed dm-verity volume and the initramfs +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + policy_name=AllowSignedAndInitramfs policy_version=0.0.0 + DEFAULT action=DENY + + op=EXECUTE boot_verified=TRUE action=ALLOW + op=EXECUTE dmverity_signature=TRUE action=ALLOW + +Prohibit execution from a specific dm-verity volume +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + policy_name=AllowSignedAndInitramfs policy_version=0.0.0 + DEFAULT action=DENY + + op=EXECUTE dmverity_roothash=sha256:cd2c5bae7c6c579edaae4353049d58eb5f2e8be0244bf05345bc8e5ed257baff action=DENY + + op=EXECUTE boot_verified=TRUE action=ALLOW + op=EXECUTE dmverity_signature=TRUE action=ALLOW + +Allow only a specific dm-verity volume +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + policy_name=AllowSignedAndInitramfs policy_version=0.0.0 + DEFAULT action=DENY + + op=EXECUTE dmverity_roothash=sha256:401fcec5944823ae12f62726e8184407a5fa9599783f030dec146938 action=ALLOW + +Allow any signed fs-verity file +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + policy_name=AllowSignedFSVerity policy_version=0.0.0 + DEFAULT action=DENY + + op=EXECUTE fsverity_signature=TRUE action=ALLOW + +Prohibit execution of a specific fs-verity file +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +:: + + policy_name=ProhibitSpecificFSVF policy_version=0.0.0 + DEFAULT action=DENY + + op=EXECUTE fsverity_digest=sha256:fd88f2b8824e197f850bf4c5109bea5cf0ee38104f710843bb72da796ba5af9e action=DENY + op=EXECUTE boot_verified=TRUE action=ALLOW + op=EXECUTE dmverity_signature=TRUE action=ALLOW + +Additional Information +---------------------- + +- `Github Repository <https://github.com/microsoft/ipe>`_ +- Documentation/security/ipe.rst + +FAQ +--- + +Q: + What's the difference between other LSMs which provide a measure of + trust-based access control? + +A: + + In general, there's two other LSMs that can provide similar functionality: + IMA, and Loadpin. + + IMA and IPE are functionally very similar. The significant difference between + the two is the policy. [#devdoc]_ + + Loadpin and IPE differ fairly dramatically, as Loadpin only covers the IPE's + kernel read operations, whereas IPE is capable of controlling execution + on top of kernel read. The trust model is also different; Loadpin roots its + trust in the initial super-block, whereas trust in IPE is stemmed from kernel + itself (via ``SYSTEM_TRUSTED_KEYS``). + +----------- + +.. [#diglim] https://lore.kernel.org/bpf/4d6932e96d774227b42721d9f645ba51@huawei.com/T/ + +.. [#interpreters] There is `some interest in solving this issue <https://lore.kernel.org/lkml/20220321161557.495388-1-mic@digikod.net/>`_. + +.. [#devdoc] Please see Documentation/security/ipe.rst for more on this topic. + +.. [#switch_root] https://man7.org/linux/man-pages/man8/switch_root.8.html + +.. [#fsveritydigest] These hash algorithms are based on values accepted by fsverity-utils; + IPE does not impose any restrictions on the digest algorithm itself; + thus, this list may be out of date. + +.. [#dmveritydigests] These hash algorithms are based on values accepted by dm-verity, + specifically ``crypto_alloc_ahash`` in ``verity_ctr``; ``veritysetup`` + does support more algorithms than the list above. IPE does not impose + any restrictions on the digest algorithm itself; thus, this list + may be out of date. + +.. [#securedigest] Please ensure you are using cryptographically secure hash functions; + just because something is *supported* does not mean it is *secure*. diff --git a/Documentation/admin-guide/kernel-parameters.txt b/Documentation/admin-guide/kernel-parameters.txt index 62feb8f31381..0767135d1111 100644 --- a/Documentation/admin-guide/kernel-parameters.txt +++ b/Documentation/admin-guide/kernel-parameters.txt @@ -2305,6 +2305,18 @@ ipcmni_extend [KNL,EARLY] Extend the maximum number of unique System V IPC identifiers from 32,768 to 16,777,216. + ipe.enforce= [IPE] + Format: <bool> + Determine whether IPE starts in permissive (0) or + enforce (1) mode. The default is enforce. + + ipe.success_audit= + [IPE] + Format: <bool> + Start IPE with success auditing enabled, emitting + an audit event when a binary is allowed. The default + is 0. + irqaffinity= [SMP] Set the default irq affinity mask The argument is a cpu list, as described above. diff --git a/Documentation/filesystems/fsverity.rst b/Documentation/filesystems/fsverity.rst index 708c79631f32..f09308099bd7 100644 --- a/Documentation/filesystems/fsverity.rst +++ b/Documentation/filesystems/fsverity.rst @@ -497,6 +497,7 @@ be carefully considered before using them: a kernel-enforced, system-wide authentication policy that allows only files with an fs-verity signature enabled to perform certain operations, such as execution. Note that IPE doesn't require fs.verity.require_signatures=1. + Please refer to Documentation/admin-guide/LSM/ipe.rst for more details. - A file's builtin signature can only be set at the same time that fs-verity is being enabled on the file. Changing or deleting the diff --git a/Documentation/security/index.rst b/Documentation/security/index.rst index 59f8fc106cb0..3e0a7114a862 100644 --- a/Documentation/security/index.rst +++ b/Documentation/security/index.rst @@ -19,3 +19,4 @@ Security Documentation digsig landlock secrets/index + ipe diff --git a/Documentation/security/ipe.rst b/Documentation/security/ipe.rst new file mode 100644 index 000000000000..ec0f41844c34 --- /dev/null +++ b/Documentation/security/ipe.rst @@ -0,0 +1,444 @@ +.. SPDX-License-Identifier: GPL-2.0 + +Integrity Policy Enforcement (IPE) - Kernel Documentation +========================================================= + +.. NOTE:: + + This is documentation targeted at developers, instead of administrators. + If you're looking for documentation on the usage of IPE, please see + Documentation/admin-guide/LSM/ipe.rst + +Historical Motivation +--------------------- + +The original issue that prompted IPE's implementation was the creation +of a locked-down system. This system would be born-secure, and have +strong integrity guarantees over both the executable code, and specific +*data files* on the system, that were critical to its function. These +specific data files would not be readable unless they passed integrity +policy. A mandatory access control system would be present, and +as a result, xattrs would have to be protected. This lead to a selection +of what would provide the integrity claims. At the time, there were two +main mechanisms considered that could guarantee integrity for the system +with these requirements: + + 1. IMA + EVM Signatures + 2. DM-Verity + +Both options were carefully considered, however the choice to use DM-Verity +over IMA+EVM as the *integrity mechanism* in the original use case of IPE +was due to three main reasons: + + 1. Protection of additional attack vectors: + + * With IMA+EVM, without an encryption solution, the system is vulnerable + to offline attack against the aforementioned specific data files. + + Unlike executables, read operations (like those on the protected data + files), cannot be enforced to be globally integrity verified. This means + there must be some form of selector to determine whether a read should + enforce the integrity policy, or it should not. + + At the time, this was done with mandatory access control labels. An IMA + policy would indicate what labels required integrity verification, which + presented an issue: EVM would protect the label, but if an attacker could + modify filesystem offline, the attacker could wipe all the xattrs - + including the SELinux labels that would be used to determine whether the + file should be subject to integrity policy. + + With DM-Verity, as the xattrs are saved as part of the merkel tree, if + offline mount occurs against the filesystem protected by dm-verity, the + checksum no longer matches and the file fails to be read. + + * As userspace binaries are paged in Linux, dm-verity also offers the + additional protection against a hostile block device. In such an attack, + the block device reports the appropriate content for the IMA hash + initially, passing the required integrity check. Then, on the page fault + that accesses the real data, will report the attacker's payload. Since + dm-verity will check the data when the page fault occurs (and the disk + access), this attack is mitigated. + + 2. Performance: + + * dm-verity provides integrity verification on demand as blocks are + read versus requiring the entire file being read into memory for + validation. + + 3. Simplicity of signing: + + * No need for two signatures (IMA, then EVM): one signature covers + an entire block device. + * Signatures can be stored externally to the filesystem metadata. + * The signature supports an x.509-based signing infrastructure. + +The next step was to choose a *policy* to enforce the integrity mechanism. +The minimum requirements for the policy were: + + 1. The policy itself must be integrity verified (preventing trivial + attack against it). + 2. The policy itself must be resistant to rollback attacks. + 3. The policy enforcement must have a permissive-like mode. + 4. The policy must be able to be updated, in its entirety, without + a reboot. + 5. Policy updates must be atomic. + 6. The policy must support *revocations* of previously authored + components. + 7. The policy must be auditable, at any point-of-time. + +IMA, as the only integrity policy mechanism at the time, was +considered against these list of requirements, and did not fulfill +all of the minimum requirements. Extending IMA to cover these +requirements was considered, but ultimately discarded for a +two reasons: + + 1. Regression risk; many of these changes would result in + dramatic code changes to IMA, which is already present in the + kernel, and therefore might impact users. + + 2. IMA was used in the system for measurement and attestation; + separation of measurement policy from local integrity policy + enforcement was considered favorable. + +Due to these reasons, it was decided that a new LSM should be created, +whose responsibility would be only the local integrity policy enforcement. + +Role and Scope +-------------- + +IPE, as its name implies, is fundamentally an integrity policy enforcement +solution; IPE does not mandate how integrity is provided, but instead +leaves that decision to the system administrator to set the security bar, +via the mechanisms that they select that suit their individual needs. +There are several different integrity solutions that provide a different +level of security guarantees; and IPE allows sysadmins to express policy for +theoretically all of them. + +IPE does not have an inherent mechanism to ensure integrity on its own. +Instead, there are more effective layers available for building systems that +can guarantee integrity. It's important to note that the mechanism for proving +integrity is independent of the policy for enforcing that integrity claim. + +Therefore, IPE was designed around: + + 1. Easy integrations with integrity providers. + 2. Ease of use for platform administrators/sysadmins. + +Design Rationale: +----------------- + +IPE was designed after evaluating existing integrity policy solutions +in other operating systems and environments. In this survey of other +implementations, there were a few pitfalls identified: + + 1. Policies were not readable by humans, usually requiring a binary + intermediary format. + 2. A single, non-customizable action was implicitly taken as a default. + 3. Debugging the policy required manual steps to determine what rule was violated. + 4. Authoring a policy required an in-depth knowledge of the larger system, + or operating system. + +IPE attempts to avoid all of these pitfalls. + +Policy +~~~~~~ + +Plain Text +^^^^^^^^^^ + +IPE's policy is plain-text. This introduces slightly larger policy files than +other LSMs, but solves two major problems that occurs with some integrity policy +solutions on other platforms. + +The first issue is one of code maintenance and duplication. To author policies, +the policy has to be some form of string representation (be it structured, +through XML, JSON, YAML, etcetera), to allow the policy author to understand +what is being written. In a hypothetical binary policy design, a serializer +is necessary to write the policy from the human readable form, to the binary +form, and a deserializer is needed to interpret the binary form into a data +structure in the kernel. + +Eventually, another deserializer will be needed to transform the binary from +back into the human-readable form with as much information preserved. This is because a +user of this access control system will have to keep a lookup table of a checksum +and the original file itself to try to understand what policies have been deployed +on this system and what policies have not. For a single user, this may be alright, +as old policies can be discarded almost immediately after the update takes hold. +For users that manage computer fleets in the thousands, if not hundreds of thousands, +with multiple different operating systems, and multiple different operational needs, +this quickly becomes an issue, as stale policies from years ago may be present, +quickly resulting in the need to recover the policy or fund extensive infrastructure +to track what each policy contains. + +With now three separate serializer/deserializers, maintenance becomes costly. If the +policy avoids the binary format, there is only one required serializer: from the +human-readable form to the data structure in kernel, saving on code maintenance, +and retaining operability. + +The second issue with a binary format is one of transparency. As IPE controls +access based on the trust of the system's resources, it's policy must also be +trusted to be changed. This is done through signatures, resulting in needing +signing as a process. Signing, as a process, is typically done with a +high security bar, as anything signed can be used to attack integrity +enforcement systems. It is also important that, when signing something, that +the signer is aware of what they are signing. A binary policy can cause +obfuscation of that fact; what signers see is an opaque binary blob. A +plain-text policy, on the other hand, the signers see the actual policy +submitted for signing. + +Boot Policy +~~~~~~~~~~~ + +IPE, if configured appropriately, is able to enforce a policy as soon as a +kernel is booted and usermode starts. That implies some level of storage +of the policy to apply the minute usermode starts. Generally, that storage +can be handled in one of three ways: + + 1. The policy file(s) live on disk and the kernel loads the policy prior + to an code path that would result in an enforcement decision. + 2. The policy file(s) are passed by the bootloader to the kernel, who + parses the policy. + 3. There is a policy file that is compiled into the kernel that is + parsed and enforced on initialization. + +The first option has problems: the kernel reading files from userspace +is typically discouraged and very uncommon in the kernel. + +The second option also has problems: Linux supports a variety of bootloaders +across its entire ecosystem - every bootloader would have to support this +new methodology or there must be an independent source. It would likely +result in more drastic changes to the kernel startup than necessary. + +The third option is the best but it's important to be aware that the policy +will take disk space against the kernel it's compiled in. It's important to +keep this policy generalized enough that userspace can load a new, more +complicated policy, but restrictive enough that it will not overauthorize +and cause security issues. + +The initramfs provides a way that this bootup path can be established. The +kernel starts with a minimal policy, that trusts the initramfs only. Inside +the initramfs, when the real rootfs is mounted, but not yet transferred to, +it deploys and activates a policy that trusts the new root filesystem. +This prevents overauthorization at any step, and keeps the kernel policy +to a minimal size. + +Startup +^^^^^^^ + +Not every system, however starts with an initramfs, so the startup policy +compiled into the kernel will need some flexibility to express how trust +is established for the next phase of the bootup. To this end, if we just +make the compiled-in policy a full IPE policy, it allows system builders +to express the first stage bootup requirements appropriately. + +Updatable, Rebootless Policy +~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +As requirements change over time (vulnerabilities are found in previously +trusted applications, keys roll, etcetera). Updating a kernel to change the +meet those security goals is not always a suitable option, as updates are not +always risk-free, and blocking a security update leaves systems vulnerable. +This means IPE requires a policy that can be completely updated (allowing +revocations of existing policy) from a source external to the kernel (allowing +policies to be updated without updating the kernel). + +Additionally, since the kernel is stateless between invocations, and reading +policy files off the disk from kernel space is a bad idea(tm), then the +policy updates have to be done rebootlessly. + +To allow an update from an external source, it could be potentially malicious, +so this policy needs to have a way to be identified as trusted. This is +done via a signature chained to a trust source in the kernel. Arbitrarily, +this is the ``SYSTEM_TRUSTED_KEYRING``, a keyring that is initially +populated at kernel compile-time, as this matches the expectation that the +author of the compiled-in policy described above is the same entity that can +deploy policy updates. + +Anti-Rollback / Anti-Replay +~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +Over time, vulnerabilities are found and trusted resources may not be +trusted anymore. IPE's policy has no exception to this. There can be +instances where a mistaken policy author deploys an insecure policy, +before correcting it with a secure policy. + +Assuming that as soon as the insecure policy is signed, and an attacker +acquires the insecure policy, IPE needs a way to prevent rollback +from the secure policy update to the insecure policy update. + +Initially, IPE's policy can have a policy_version that states the +minimum required version across all policies that can be active on +the system. This will prevent rollback while the system is live. + +.. WARNING:: + + However, since the kernel is stateless across boots, this policy + version will be reset to 0.0.0 on the next boot. System builders + need to be aware of this, and ensure the new secure policies are + deployed ASAP after a boot to ensure that the window of + opportunity is minimal for an attacker to deploy the insecure policy. + +Implicit Actions: +~~~~~~~~~~~~~~~~~ + +The issue of implicit actions only becomes visible when you consider +a mixed level of security bars across multiple operations in a system. +For example, consider a system that has strong integrity guarantees +over both the executable code, and specific *data files* on the system, +that were critical to its function. In this system, three types of policies +are possible: + + 1. A policy in which failure to match any rules in the policy results + in the action being denied. + 2. A policy in which failure to match any rules in the policy results + in the action being allowed. + 3. A policy in which the action taken when no rules are matched is + specified by the policy author. + +The first option could make a policy like this:: + + op=EXECUTE integrity_verified=YES action=ALLOW + +In the example system, this works well for the executables, as all +executables should have integrity guarantees, without exception. The +issue becomes with the second requirement about specific data files. +This would result in a policy like this (assuming each line is +evaluated in order):: + + op=EXECUTE integrity_verified=YES action=ALLOW + + op=READ integrity_verified=NO label=critical_t action=DENY + op=READ action=ALLOW + +This is somewhat clear if you read the docs, understand the policy +is executed in order and that the default is a denial; however, the +last line effectively changes that default to an ALLOW. This is +required, because in a realistic system, there are some unverified +reads (imagine appending to a log file). + +The second option, matching no rules results in an allow, is clearer +for the specific data files:: + + op=READ integrity_verified=NO label=critical_t action=DENY + +And, like the first option, falls short with the opposite scenario, +effectively needing to override the default:: + + op=EXECUTE integrity_verified=YES action=ALLOW + op=EXECUTE action=DENY + + op=READ integrity_verified=NO label=critical_t action=DENY + +This leaves the third option. Instead of making users be clever +and override the default with an empty rule, force the end-user +to consider what the appropriate default should be for their +scenario and explicitly state it:: + + DEFAULT op=EXECUTE action=DENY + op=EXECUTE integrity_verified=YES action=ALLOW + + DEFAULT op=READ action=ALLOW + op=READ integrity_verified=NO label=critical_t action=DENY + +Policy Debugging: +~~~~~~~~~~~~~~~~~ + +When developing a policy, it is useful to know what line of the policy +is being violated to reduce debugging costs; narrowing the scope of the +investigation to the exact line that resulted in the action. Some integrity +policy systems do not provide this information, instead providing the +information that was used in the evaluation. This then requires a correlation +with the policy to evaluate what went wrong. + +Instead, IPE just emits the rule that was matched. This limits the scope +of the investigation to the exact policy line (in the case of a specific +rule), or the section (in the case of a DEFAULT). This decreases iteration +and investigation times when policy failures are observed while evaluating +policies. + +IPE's policy engine is also designed in a way that it makes it obvious to +a human of how to investigate a policy failure. Each line is evaluated in +the sequence that is written, so the algorithm is very simple to follow +for humans to recreate the steps and could have caused the failure. In other +surveyed systems, optimizations occur (sorting rules, for instance) when loading +the policy. In those systems, it requires multiple steps to debug, and the +algorithm may not always be clear to the end-user without reading the code first. + +Simplified Policy: +~~~~~~~~~~~~~~~~~~ + +Finally, IPE's policy is designed for sysadmins, not kernel developers. Instead +of covering individual LSM hooks (or syscalls), IPE covers operations. This means +instead of sysadmins needing to know that the syscalls ``mmap``, ``mprotect``, +``execve``, and ``uselib`` must have rules protecting them, they must simple know +that they want to restrict code execution. This limits the amount of bypasses that +could occur due to a lack of knowledge of the underlying system; whereas the +maintainers of IPE, being kernel developers can make the correct choice to determine +whether something maps to these operations, and under what conditions. + +Implementation Notes +-------------------- + +Anonymous Memory +~~~~~~~~~~~~~~~~ + +Anonymous memory isn't treated any differently from any other access in IPE. +When anonymous memory is mapped with ``+X``, it still comes into the ``file_mmap`` +or ``file_mprotect`` hook, but with a ``NULL`` file object. This is submitted to +the evaluation, like any other file, however, all current trust mechanisms will +return false as there is nothing to evaluate. This means anonymous memory +execution is subject to whatever the ``DEFAULT`` is for ``EXECUTE``. + +.. WARNING:: + + This also occurs with the ``kernel_load_data`` hook, which is used by signed + and compressed kernel modules. Using signed and compressed kernel modules with + IPE will always result in the ``DEFAULT`` action for ``KMODULE``. + +Securityfs Interface +~~~~~~~~~~~~~~~~~~~~ + +The per-policy securityfs tree is somewhat unique. For example, for +a standard securityfs policy tree:: + + MyPolicy + |- active + |- delete + |- name + |- pkcs7 + |- policy + |- update + |- version + +The policy is stored in the ``->i_private`` data of the MyPolicy inode. + +Tests +----- + +IPE has KUnit Tests for the policy parser. Recommended kunitconfig:: + + CONFIG_KUNIT=y + CONFIG_SECURITY=y + CONFIG_SECURITYFS=y + CONFIG_PKCS7_MESSAGE_PARSER=y + CONFIG_SYSTEM_DATA_VERIFICATION=y + CONFIG_FS_VERITY=y + CONFIG_FS_VERITY_BUILTIN_SIGNATURES=y + CONFIG_BLOCK=y + CONFIG_MD=y + CONFIG_BLK_DEV_DM=y + CONFIG_DM_VERITY=y + CONFIG_DM_VERITY_VERIFY_ROOTHASH_SIG=y + CONFIG_NET=y + CONFIG_AUDIT=y + CONFIG_AUDITSYSCALL=y + CONFIG_BLK_DEV_INITRD=y + + CONFIG_SECURITY_IPE=y + CONFIG_IPE_PROP_DM_VERITY=y + CONFIG_IPE_PROP_FS_VERITY=y + CONFIG_SECURITY_IPE_KUNIT_TEST=y + +In addition, IPE has a python based integration +`test suite <https://github.com/microsoft/ipe/tree/test-suite>`_ that +can test both user interfaces and enforcement functionalities. diff --git a/MAINTAINERS b/MAINTAINERS index d84739ab2ef4..ee081f7d5fa3 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -10715,6 +10715,8 @@ M: Fan Wu <wufan@linux.microsoft.com> L: linux-security-module@vger.kernel.org S: Supported T: git https://github.com/microsoft/ipe.git +F: Documentation/admin-guide/LSM/ipe.rst +F: Documentation/security/ipe.rst F: scripts/ipe/ F: security/ipe/ -- 2.44.0
From: Deven Bowers <deven.desai@linux.microsoft.com> Add various happy/unhappy unit tests for both IPE's policy parser. Besides, a test suite for IPE functionality is available at https://github.com/microsoft/ipe/tree/test-suite Signed-off-by: Deven Bowers <deven.desai@linux.microsoft.com> Signed-off-by: Fan Wu <wufan@linux.microsoft.com> --- v1-v6: + Not present v7: Introduced v8: + Remove the kunit tests with respect to the fsverity digest, as these require significant changes to work with the new method of acquiring the digest at runtime. v9: + Remove the kunit tests related to ipe_context v10: + No changes v11: + No changes v12: + No changes v13: + No changes v14: + No changes v15: + No changes --- security/ipe/Kconfig | 17 +++ security/ipe/Makefile | 3 + security/ipe/policy_tests.c | 294 ++++++++++++++++++++++++++++++++++++ 3 files changed, 314 insertions(+) create mode 100644 security/ipe/policy_tests.c diff --git a/security/ipe/Kconfig b/security/ipe/Kconfig index a18c5915b723..1b2a306155ae 100644 --- a/security/ipe/Kconfig +++ b/security/ipe/Kconfig @@ -55,4 +55,21 @@ config IPE_PROP_FS_VERITY endmenu +config SECURITY_IPE_KUNIT_TEST + bool "Build KUnit tests for IPE" if !KUNIT_ALL_TESTS + depends on KUNIT=y + default KUNIT_ALL_TESTS + help + This builds the IPE KUnit tests. + + KUnit tests run during boot and output the results to the debug log + in TAP format (https://testanything.org/). Only useful for kernel devs + running KUnit test harness and are not for inclusion into a + production build. + + For more information on KUnit and unit tests in general please refer + to the KUnit documentation in Documentation/dev-tools/kunit/. + + If unsure, say N. + endif diff --git a/security/ipe/Makefile b/security/ipe/Makefile index ce23101b66ba..def674e09e84 100644 --- a/security/ipe/Makefile +++ b/security/ipe/Makefile @@ -26,3 +26,6 @@ obj-$(CONFIG_SECURITY_IPE) += \ audit.o \ clean-files := boot-policy.c \ + +obj-$(CONFIG_SECURITY_IPE_KUNIT_TEST) += \ + policy_tests.o \ diff --git a/security/ipe/policy_tests.c b/security/ipe/policy_tests.c new file mode 100644 index 000000000000..d67c42c9ca59 --- /dev/null +++ b/security/ipe/policy_tests.c @@ -0,0 +1,294 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) Microsoft Corporation. All rights reserved. + */ + +#include <linux/slab.h> +#include <linux/types.h> +#include <linux/list.h> +#include <kunit/test.h> +#include "policy.h" +struct policy_case { + const char *const policy; + int errno; + const char *const desc; +}; + +static const struct policy_case policy_cases[] = { + { + "policy_name=allowall policy_version=0.0.0\n" + "DEFAULT action=ALLOW", + 0, + "basic", + }, + { + "policy_name=trailing_comment policy_version=152.0.0 #This is comment\n" + "DEFAULT action=ALLOW", + 0, + "trailing comment", + }, + { + "policy_name=allowallnewline policy_version=0.2.0\n" + "DEFAULT action=ALLOW\n" + "\n", + 0, + "trailing newline", + }, + { + "policy_name=carriagereturnlinefeed policy_version=0.0.1\n" + "DEFAULT action=ALLOW\n" + "\r\n", + 0, + "clrf newline", + }, + { + "policy_name=whitespace policy_version=0.0.0\n" + "DEFAULT\taction=ALLOW\n" + " \t DEFAULT \t op=EXECUTE action=DENY\n" + "op=EXECUTE boot_verified=TRUE action=ALLOW\n" + "# this is a\tcomment\t\t\t\t\n" + "DEFAULT \t op=KMODULE\t\t\t action=DENY\r\n" + "op=KMODULE boot_verified=TRUE action=ALLOW\n", + 0, + "various whitespaces and nested default", + }, + { + "policy_name=boot_verified policy_version=-1236.0.0\n" + "DEFAULT\taction=ALLOW\n", + -EINVAL, + "negative version", + }, + { + "policy_name=$@!*&^%%\\:;{}() policy_version=0.0.0\n" + "DEFAULT action=ALLOW", + 0, + "special characters", + }, + { + "policy_name=test policy_version=999999.0.0\n" + "DEFAULT action=ALLOW", + -ERANGE, + "overflow version", + }, + { + "policy_name=test policy_version=255.0\n" + "DEFAULT action=ALLOW", + -EBADMSG, + "incomplete version", + }, + { + "policy_name=test policy_version=111.0.0.0\n" + "DEFAULT action=ALLOW", + -EBADMSG, + "extra version", + }, + { + "", + -EBADMSG, + "0-length policy", + }, + { + "policy_name=test\0policy_version=0.0.0\n" + "DEFAULT action=ALLOW", + -EBADMSG, + "random null in header", + }, + { + "policy_name=test policy_version=0.0.0\n" + "\0DEFAULT action=ALLOW", + -EBADMSG, + "incomplete policy from NULL", + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=DENY\n\0" + "op=EXECUTE dmverity_signature=TRUE action=ALLOW\n", + 0, + "NULL truncates policy", + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "op=EXECUTE dmverity_signature=abc action=ALLOW", + -EBADMSG, + "invalid property type", + }, + { + "DEFAULT action=ALLOW", + -EBADMSG, + "missing policy header", + }, + { + "policy_name=test policy_version=0.0.0\n", + -EBADMSG, + "missing default definition", + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "dmverity_signature=TRUE op=EXECUTE action=ALLOW", + -EBADMSG, + "invalid rule ordering" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "action=ALLOW op=EXECUTE dmverity_signature=TRUE", + -EBADMSG, + "invalid rule ordering (2)", + }, + { + "policy_name=test policy_version=0.0\n" + "DEFAULT action=ALLOW\n" + "op=EXECUTE dmverity_signature=TRUE action=ALLOW", + -EBADMSG, + "invalid version", + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "op=UNKNOWN dmverity_signature=TRUE action=ALLOW", + -EBADMSG, + "unknown operation", + }, + { + "policy_name=asdvpolicy_version=0.0.0\n" + "DEFAULT action=ALLOW\n", + -EBADMSG, + "missing space after policy name", + }, + { + "policy_name=test\xFF\xEF policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "op=EXECUTE dmverity_signature=TRUE action=ALLOW", + 0, + "expanded ascii", + }, + { + "policy_name=test\xFF\xEF policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "op=EXECUTE dmverity_roothash=GOOD_DOG action=ALLOW", + -EBADMSG, + "invalid property value (2)", + }, + { + "policy_name=test policy_version=0.0.0\n" + "policy_name=test policy_version=0.1.0\n" + "DEFAULT action=ALLOW", + -EBADMSG, + "double header" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "DEFAULT action=ALLOW\n", + -EBADMSG, + "double default" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "DEFAULT op=EXECUTE action=DENY\n" + "DEFAULT op=EXECUTE action=ALLOW\n", + -EBADMSG, + "double operation default" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "DEFAULT op=EXECUTE action=DEN\n", + -EBADMSG, + "invalid action value" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "DEFAULT op=EXECUTE action\n", + -EBADMSG, + "invalid action value (2)" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "UNKNOWN value=true\n", + -EBADMSG, + "unrecognized statement" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "op=EXECUTE dmverity_roothash=1c0d7ee1f8343b7fbe418378e8eb22c061d7dec7 action=DENY\n", + -EBADMSG, + "old-style digest" + }, + { + "policy_name=test policy_version=0.0.0\n" + "DEFAULT action=ALLOW\n" + "op=EXECUTE fsverity_digest=1c0d7ee1f8343b7fbe418378e8eb22c061d7dec7 action=DENY\n", + -EBADMSG, + "old-style digest" + } +}; + +static void pol_to_desc(const struct policy_case *c, char *desc) +{ + strscpy(desc, c->desc, KUNIT_PARAM_DESC_SIZE); +} + +KUNIT_ARRAY_PARAM(ipe_policies, policy_cases, pol_to_desc); + +/** + * ipe_parser_unsigned_test - Test the parser by passing unsigned policies. + * @test: Supplies a pointer to a kunit structure. + * + * This is called by the kunit harness. This test does not check the correctness + * of the policy, but ensures that errors are handled correctly. + */ +static void ipe_parser_unsigned_test(struct kunit *test) +{ + const struct policy_case *p = test->param_value; + struct ipe_policy *pol = ipe_new_policy(p->policy, strlen(p->policy), NULL, 0); + + if (p->errno) { + KUNIT_EXPECT_EQ(test, PTR_ERR(pol), p->errno); + return; + } + + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pol); + KUNIT_EXPECT_NOT_ERR_OR_NULL(test, pol->parsed); + KUNIT_EXPECT_STREQ(test, pol->text, p->policy); + KUNIT_EXPECT_PTR_EQ(test, NULL, pol->pkcs7); + KUNIT_EXPECT_EQ(test, 0, pol->pkcs7len); + + ipe_free_policy(pol); +} + +/** + * ipe_parser_widestring_test - Ensure parser fail on a wide string policy. + * @test: Supplies a pointer to a kunit structure. + * + * This is called by the kunit harness. + */ +static void ipe_parser_widestring_test(struct kunit *test) +{ + struct ipe_policy *pol = NULL; + const unsigned short policy[] = L"policy_name=Test policy_version=0.0.0\n" + L"DEFAULT action=ALLOW"; + + pol = ipe_new_policy((const char *)policy, (ARRAY_SIZE(policy) - 1) * 2, NULL, 0); + KUNIT_EXPECT_TRUE(test, IS_ERR_OR_NULL(pol)); + + ipe_free_policy(pol); +} + +static struct kunit_case ipe_parser_test_cases[] = { + KUNIT_CASE_PARAM(ipe_parser_unsigned_test, ipe_policies_gen_params), + KUNIT_CASE(ipe_parser_widestring_test), +}; + +static struct kunit_suite ipe_parser_test_suite = { + .name = "ipe-parser", + .test_cases = ipe_parser_test_cases, +}; + +kunit_test_suite(ipe_parser_test_suite); -- 2.44.0
From: Deven Bowers <deven.desai@linux.microsoft.com> Allows author of IPE policy to indicate trust for a singular dm-verity volume, identified by roothash, through "dmverity_roothash" and all signed dm-verity volumes, through "dmverity_signature". Signed-off-by: Deven Bowers <deven.desai@linux.microsoft.com> Signed-off-by: Fan Wu <wufan@linux.microsoft.com> --- v2: + No Changes v3: + No changes v4: + No changes v5: + No changes v6: + Fix an improper cleanup that can result in a leak v7: + Squash patch 08/12, 10/12 to [11/16] v8: + Undo squash of 08/12, 10/12 - separating drivers/md/ from security/ & block/ + Use common-audit function for dmverity_signature. + Change implementation for storing the dm-verity digest to use the newly introduced dm_verity_digest structure introduced in patch 14/20. v9: + Adapt to the new parser v10: + Select the Kconfig when all dependencies are enabled v11: + No changes v12: + Refactor to use struct digest_info* instead of void* + Correct audit format v13: + Remove the CONFIG_IPE_PROP_DM_VERITY dependency inside the parser to make the policy grammar independent of the kernel config. v14: + No changes v15: + Fix one grammar issue in KCONFIG + Switch to use security_bdev_setintegrity() hook --- security/ipe/Kconfig | 18 ++++++ security/ipe/Makefile | 1 + security/ipe/audit.c | 29 ++++++++- security/ipe/digest.c | 120 +++++++++++++++++++++++++++++++++++ security/ipe/digest.h | 26 ++++++++ security/ipe/eval.c | 91 +++++++++++++++++++++++++- security/ipe/eval.h | 10 +++ security/ipe/hooks.c | 67 +++++++++++++++++++ security/ipe/hooks.h | 8 +++ security/ipe/ipe.c | 15 +++++ security/ipe/ipe.h | 4 ++ security/ipe/policy.h | 3 + security/ipe/policy_parser.c | 24 ++++++- 13 files changed, 412 insertions(+), 4 deletions(-) create mode 100644 security/ipe/digest.c create mode 100644 security/ipe/digest.h diff --git a/security/ipe/Kconfig b/security/ipe/Kconfig index ac4d558e69d5..5433ca3fc855 100644 --- a/security/ipe/Kconfig +++ b/security/ipe/Kconfig @@ -8,6 +8,7 @@ menuconfig SECURITY_IPE depends on SECURITY && SECURITYFS && AUDIT && AUDITSYSCALL select PKCS7_MESSAGE_PARSER select SYSTEM_DATA_VERIFICATION + select IPE_PROP_DM_VERITY if DM_VERITY && DM_VERITY_VERIFY_ROOTHASH_SIG help This option enables the Integrity Policy Enforcement LSM allowing users to define a policy to enforce a trust-based access @@ -15,3 +16,20 @@ menuconfig SECURITY_IPE admins to reconfigure trust requirements on the fly. If unsure, answer N. + +if SECURITY_IPE +menu "IPE Trust Providers" + +config IPE_PROP_DM_VERITY + bool "Enable support for dm-verity volumes" + depends on DM_VERITY && DM_VERITY_VERIFY_ROOTHASH_SIG + help + This option enables the properties 'dmverity_signature' and + 'dmverity_roothash' in IPE policy. These properties evaluate + to TRUE when a file is evaluated against a dm-verity volume + that was mounted with a signed root-hash or the volume's + root hash matches the supplied value in the policy. + +endmenu + +endif diff --git a/security/ipe/Makefile b/security/ipe/Makefile index 2279eaa3cea3..66de53687d11 100644 --- a/security/ipe/Makefile +++ b/security/ipe/Makefile @@ -6,6 +6,7 @@ # obj-$(CONFIG_SECURITY_IPE) += \ + digest.o \ eval.o \ hooks.o \ fs.o \ diff --git a/security/ipe/audit.c b/security/ipe/audit.c index a480f27278df..3acbfecc9f43 100644 --- a/security/ipe/audit.c +++ b/security/ipe/audit.c @@ -13,6 +13,7 @@ #include "hooks.h" #include "policy.h" #include "audit.h" +#include "digest.h" #define ACTSTR(x) ((x) == IPE_ACTION_ALLOW ? "ALLOW" : "DENY") @@ -49,8 +50,22 @@ static const char *const audit_hook_names[__IPE_HOOK_MAX] = { static const char *const audit_prop_names[__IPE_PROP_MAX] = { "boot_verified=FALSE", "boot_verified=TRUE", + "dmverity_roothash=", + "dmverity_signature=FALSE", + "dmverity_signature=TRUE", }; +/** + * audit_dmv_roothash - audit the roothash of a dmverity_roothash property. + * @ab: Supplies a pointer to the audit_buffer to append to. + * @rh: Supplies a pointer to the digest structure. + */ +static void audit_dmv_roothash(struct audit_buffer *ab, const void *rh) +{ + audit_log_format(ab, "%s", audit_prop_names[IPE_PROP_DMV_ROOTHASH]); + ipe_digest_audit(ab, rh); +} + /** * audit_rule - audit an IPE policy rule approximation. * @ab: Supplies a pointer to the audit_buffer to append to. @@ -62,8 +77,18 @@ static void audit_rule(struct audit_buffer *ab, const struct ipe_rule *r) audit_log_format(ab, " rule=\"%s ", audit_op_names[r->op]); - list_for_each_entry(ptr, &r->props, next) - audit_log_format(ab, "%s ", audit_prop_names[ptr->type]); + list_for_each_entry(ptr, &r->props, next) { + switch (ptr->type) { + case IPE_PROP_DMV_ROOTHASH: + audit_dmv_roothash(ab, ptr->value); + break; + default: + audit_log_format(ab, "%s", audit_prop_names[ptr->type]); + break; + } + + audit_log_format(ab, " "); + } audit_log_format(ab, "action=%s\"", ACTSTR(r->action)); } diff --git a/security/ipe/digest.c b/security/ipe/digest.c new file mode 100644 index 000000000000..f6fde0b1fe74 --- /dev/null +++ b/security/ipe/digest.c @@ -0,0 +1,120 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) Microsoft Corporation. All rights reserved. + */ + +#include "digest.h" + +/** + * ipe_digest_parse - parse a digest in IPE's policy. + * @valstr: Supplies the string parsed from the policy. + * + * Digests in IPE are defined in a standard way: + * <alg_name>:<hex> + * + * Use this function to create a property to parse the digest + * consistently. The parsed digest will be saved in @value in IPE's + * policy. + * + * Return: The parsed digest_info structure. + */ +struct digest_info *ipe_digest_parse(const char *valstr) +{ + char *sep, *raw_digest; + size_t raw_digest_len; + int rc = 0; + char *alg = NULL; + u8 *digest = NULL; + struct digest_info *info = NULL; + + info = kzalloc(sizeof(*info), GFP_KERNEL); + if (!info) + return ERR_PTR(-ENOMEM); + + sep = strchr(valstr, ':'); + if (!sep) { + rc = -EBADMSG; + goto err; + } + + alg = kstrndup(valstr, sep - valstr, GFP_KERNEL); + if (!alg) { + rc = -ENOMEM; + goto err; + } + + raw_digest = sep + 1; + raw_digest_len = strlen(raw_digest); + + info->digest_len = (raw_digest_len + 1) / 2; + digest = kzalloc(info->digest_len, GFP_KERNEL); + if (!digest) { + rc = -ENOMEM; + goto err; + } + + rc = hex2bin(digest, raw_digest, info->digest_len); + if (rc < 0) { + rc = -EINVAL; + goto err; + } + + info->alg = alg; + info->digest = digest; + return info; + +err: + kfree(alg); + kfree(digest); + kfree(info); + return ERR_PTR(rc); +} + +/** + * ipe_digest_eval - evaluate an IPE digest against another digest. + * @expected: Supplies the policy-provided digest value. + * @digest: Supplies the digest to compare against the policy digest value. + * + * Return: + * * true - digests match + * * false - digests do not match + */ +bool ipe_digest_eval(const struct digest_info *expected, + const struct digest_info *digest) +{ + return (expected->digest_len == digest->digest_len) && + (!strcmp(expected->alg, digest->alg)) && + (!memcmp(expected->digest, digest->digest, expected->digest_len)); +} + +/** + * ipe_digest_free - free an IPE digest. + * @info: Supplies a pointer the policy-provided digest to free. + */ +void ipe_digest_free(struct digest_info *info) +{ + if (IS_ERR_OR_NULL(info)) + return; + + kfree(info->alg); + kfree(info->digest); + kfree(info); +} + +/** + * ipe_digest_audit - audit a digest that was sourced from IPE's policy. + * @ab: Supplies the audit_buffer to append the formatted result. + * @info: Supplies a pointer to source the audit record from. + * + * Digests in IPE are defined in a standard way: + * <alg_name>:<hex> + * + * Use this function to create a property to audit the digest + * consistently. + */ +void ipe_digest_audit(struct audit_buffer *ab, const struct digest_info *info) +{ + audit_log_untrustedstring(ab, info->alg); + audit_log_format(ab, ":"); + audit_log_n_hex(ab, info->digest, info->digest_len); +} diff --git a/security/ipe/digest.h b/security/ipe/digest.h new file mode 100644 index 000000000000..13fa67071805 --- /dev/null +++ b/security/ipe/digest.h @@ -0,0 +1,26 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Copyright (C) Microsoft Corporation. All rights reserved. + */ + +#ifndef _IPE_DIGEST_H +#define _IPE_DIGEST_H + +#include <linux/types.h> +#include <linux/audit.h> + +#include "policy.h" + +struct digest_info { + const char *alg; + const u8 *digest; + size_t digest_len; +}; + +struct digest_info *ipe_digest_parse(const char *valstr); +void ipe_digest_free(struct digest_info *digest_info); +void ipe_digest_audit(struct audit_buffer *ab, const struct digest_info *val); +bool ipe_digest_eval(const struct digest_info *expected, + const struct digest_info *digest); + +#endif /* _IPE_DIGEST_H */ diff --git a/security/ipe/eval.c b/security/ipe/eval.c index 3f3125478dd0..8dcf7809c9fc 100644 --- a/security/ipe/eval.c +++ b/security/ipe/eval.c @@ -15,10 +15,12 @@ #include "eval.h" #include "policy.h" #include "audit.h" +#include "digest.h" struct ipe_policy __rcu *ipe_active_policy; bool success_audit; bool enforce = true; +#define INO_BLOCK_DEV(ino) ((ino)->i_sb->s_bdev) #define FILE_SUPERBLOCK(f) ((f)->f_path.mnt->mnt_sb) @@ -32,6 +34,23 @@ static void build_ipe_sb_ctx(struct ipe_eval_ctx *ctx, const struct file *const ctx->initramfs = ipe_sb(FILE_SUPERBLOCK(file))->initramfs; } +#ifdef CONFIG_IPE_PROP_DM_VERITY +/** + * build_ipe_bdev_ctx - Build ipe_bdev field of an evaluation context. + * @ctx: Supplies a pointer to the context to be populated. + * @ino: Supplies the inode struct of the file triggered IPE event. + */ +static void build_ipe_bdev_ctx(struct ipe_eval_ctx *ctx, const struct inode *const ino) +{ + if (INO_BLOCK_DEV(ino)) + ctx->ipe_bdev = ipe_bdev(INO_BLOCK_DEV(ino)); +} +#else +static void build_ipe_bdev_ctx(struct ipe_eval_ctx *ctx, const struct inode *const ino) +{ +} +#endif /* CONFIG_IPE_PROP_DM_VERITY */ + /** * build_eval_ctx - Build an evaluation context. * @ctx: Supplies a pointer to the context to be populated. @@ -48,8 +67,10 @@ void build_eval_ctx(struct ipe_eval_ctx *ctx, ctx->op = op; ctx->hook = hook; - if (file) + if (file) { build_ipe_sb_ctx(ctx, file); + build_ipe_bdev_ctx(ctx, d_real_inode(file->f_path.dentry)); + } } /** @@ -65,6 +86,68 @@ static bool evaluate_boot_verified(const struct ipe_eval_ctx *const ctx) return ctx->initramfs; } +#ifdef CONFIG_IPE_PROP_DM_VERITY +/** + * evaluate_dmv_roothash - Evaluate @ctx against a dmv roothash property. + * @ctx: Supplies a pointer to the context being evaluated. + * @p: Supplies a pointer to the property being evaluated. + * + * Return: + * * true - The current @ctx match the @p + * * false - The current @ctx doesn't match the @p + */ +static bool evaluate_dmv_roothash(const struct ipe_eval_ctx *const ctx, + struct ipe_prop *p) +{ + return !!ctx->ipe_bdev && + !!ctx->ipe_bdev->root_hash && + ipe_digest_eval(p->value, + ctx->ipe_bdev->root_hash); +} + +/** + * evaluate_dmv_sig_false: Analyze @ctx against a dmv sig false property. + * @ctx: Supplies a pointer to the context being evaluated. + * + * Return: + * * true - The current @ctx match the property + * * false - The current @ctx doesn't match the property + */ +static bool evaluate_dmv_sig_false(const struct ipe_eval_ctx *const ctx) +{ + return !ctx->ipe_bdev || (!ctx->ipe_bdev->dm_verity_signed); +} + +/** + * evaluate_dmv_sig_true: Analyze @ctx against a dmv sig true property. + * @ctx: Supplies a pointer to the context being evaluated. + * + * Return: + * * true - The current @ctx match the property + * * false - The current @ctx doesn't match the property + */ +static bool evaluate_dmv_sig_true(const struct ipe_eval_ctx *const ctx) +{ + return !evaluate_dmv_sig_false(ctx); +} +#else +static bool evaluate_dmv_roothash(const struct ipe_eval_ctx *const ctx, + struct ipe_prop *p) +{ + return false; +} + +static bool evaluate_dmv_sig_false(const struct ipe_eval_ctx *const ctx) +{ + return false; +} + +static bool evaluate_dmv_sig_true(const struct ipe_eval_ctx *const ctx) +{ + return false; +} +#endif /* CONFIG_IPE_PROP_DM_VERITY */ + /** * evaluate_property - Analyze @ctx against a property. * @ctx: Supplies a pointer to the context to be evaluated. @@ -82,6 +165,12 @@ static bool evaluate_property(const struct ipe_eval_ctx *const ctx, return !evaluate_boot_verified(ctx); case IPE_PROP_BOOT_VERIFIED_TRUE: return evaluate_boot_verified(ctx); + case IPE_PROP_DMV_ROOTHASH: + return evaluate_dmv_roothash(ctx, p); + case IPE_PROP_DMV_SIG_FALSE: + return evaluate_dmv_sig_false(ctx); + case IPE_PROP_DMV_SIG_TRUE: + return evaluate_dmv_sig_true(ctx); default: return false; } diff --git a/security/ipe/eval.h b/security/ipe/eval.h index 0b50302611a4..494c0754f512 100644 --- a/security/ipe/eval.h +++ b/security/ipe/eval.h @@ -22,12 +22,22 @@ struct ipe_superblock { bool initramfs; }; +#ifdef CONFIG_IPE_PROP_DM_VERITY +struct ipe_bdev { + bool dm_verity_signed; + struct digest_info *root_hash; +}; +#endif /* CONFIG_IPE_PROP_DM_VERITY */ + struct ipe_eval_ctx { enum ipe_op_type op; enum ipe_hook_type hook; const struct file *file; bool initramfs; +#ifdef CONFIG_IPE_PROP_DM_VERITY + const struct ipe_bdev *ipe_bdev; +#endif /* CONFIG_IPE_PROP_DM_VERITY */ }; enum ipe_match { diff --git a/security/ipe/hooks.c b/security/ipe/hooks.c index 6bcc7908ed13..1abddd6e1e7f 100644 --- a/security/ipe/hooks.c +++ b/security/ipe/hooks.c @@ -8,10 +8,14 @@ #include <linux/types.h> #include <linux/binfmts.h> #include <linux/mman.h> +#include <linux/blk_types.h> +#include <linux/dm-verity.h> +#include <crypto/hash_info.h> #include "ipe.h" #include "hooks.h" #include "eval.h" +#include "digest.h" /** * ipe_bprm_check_security - ipe security hook function for bprm check. @@ -187,3 +191,66 @@ void ipe_unpack_initramfs(void) { ipe_sb(current->fs->root.mnt->mnt_sb)->initramfs = true; } + +#ifdef CONFIG_IPE_PROP_DM_VERITY +/** + * ipe_bdev_free_security - free IPE's LSM blob of block_devices. + * @bdev: Supplies a pointer to a block_device that contains the structure + * to free. + */ +void ipe_bdev_free_security(struct block_device *bdev) +{ + struct ipe_bdev *blob = ipe_bdev(bdev); + + ipe_digest_free(blob->root_hash); +} + +/** + * ipe_bdev_setintegrity - save integrity data from a bdev to IPE's LSM blob. + * @bdev: Supplies a pointer to a block_device that contains the LSM blob. + * @type: Supplies the integrity type. + * @value: Supplies the value to store. + * @len: The length of @value. + */ +int ipe_bdev_setintegrity(struct block_device *bdev, enum lsm_intgr_type type, + const void *value, size_t len) +{ + struct ipe_bdev *blob = ipe_bdev(bdev); + + if (type == LSM_INTGR_DMV_ROOTHASH) { + const struct dm_verity_digest *digest = value; + struct digest_info *info = NULL; + u8 *raw_digest = NULL; + char *alg = NULL; + + info = kzalloc(sizeof(*info), GFP_KERNEL); + if (!info) + return -ENOMEM; + + raw_digest = kmemdup(digest->digest, digest->digest_len, + GFP_KERNEL); + if (!raw_digest) + goto err; + + alg = kstrdup(digest->alg, GFP_KERNEL); + if (!alg) + goto err; + + info->alg = alg; + info->digest = raw_digest; + info->digest_len = digest->digest_len; + blob->root_hash = info; + return 0; +err: + kfree(info); + kfree(raw_digest); + kfree(alg); + return -ENOMEM; + } else if (type == LSM_INTGR_DMV_SIG) { + blob->dm_verity_signed = true; + return 0; + } + + return 0; +} +#endif /* CONFIG_IPE_PROP_DM_VERITY */ diff --git a/security/ipe/hooks.h b/security/ipe/hooks.h index b92c1d78e253..6b735aac4028 100644 --- a/security/ipe/hooks.h +++ b/security/ipe/hooks.h @@ -8,6 +8,7 @@ #include <linux/fs.h> #include <linux/binfmts.h> #include <linux/security.h> +#include <linux/blk_types.h> enum ipe_hook_type { IPE_HOOK_BPRM_CHECK = 0, @@ -35,4 +36,11 @@ int ipe_kernel_load_data(enum kernel_load_data_id id, bool contents); void ipe_unpack_initramfs(void); +#ifdef CONFIG_IPE_PROP_DM_VERITY +void ipe_bdev_free_security(struct block_device *bdev); + +int ipe_bdev_setintegrity(struct block_device *bdev, enum lsm_intgr_type type, + const void *value, size_t len); +#endif /* CONFIG_IPE_PROP_DM_VERITY */ + #endif /* _IPE_HOOKS_H */ diff --git a/security/ipe/ipe.c b/security/ipe/ipe.c index 53820ca6d5b7..3d63e9672f12 100644 --- a/security/ipe/ipe.c +++ b/security/ipe/ipe.c @@ -7,11 +7,15 @@ #include "ipe.h" #include "eval.h" #include "hooks.h" +#include "eval.h" bool ipe_enabled; static struct lsm_blob_sizes ipe_blobs __ro_after_init = { .lbs_superblock = sizeof(struct ipe_superblock), +#ifdef CONFIG_IPE_PROP_DM_VERITY + .lbs_bdev = sizeof(struct ipe_bdev), +#endif /* CONFIG_IPE_PROP_DM_VERITY */ }; static const struct lsm_id ipe_lsmid = { @@ -24,6 +28,13 @@ struct ipe_superblock *ipe_sb(const struct super_block *sb) return sb->s_security + ipe_blobs.lbs_superblock; } +#ifdef CONFIG_IPE_PROP_DM_VERITY +struct ipe_bdev *ipe_bdev(struct block_device *b) +{ + return b->security + ipe_blobs.lbs_bdev; +} +#endif /* CONFIG_IPE_PROP_DM_VERITY */ + static struct security_hook_list ipe_hooks[] __ro_after_init = { LSM_HOOK_INIT(bprm_check_security, ipe_bprm_check_security), LSM_HOOK_INIT(mmap_file, ipe_mmap_file), @@ -31,6 +42,10 @@ static struct security_hook_list ipe_hooks[] __ro_after_init = { LSM_HOOK_INIT(kernel_read_file, ipe_kernel_read_file), LSM_HOOK_INIT(kernel_load_data, ipe_kernel_load_data), LSM_HOOK_INIT(initramfs_populated, ipe_unpack_initramfs), +#ifdef CONFIG_IPE_PROP_DM_VERITY + LSM_HOOK_INIT(bdev_free_security, ipe_bdev_free_security), + LSM_HOOK_INIT(bdev_setintegrity, ipe_bdev_setintegrity), +#endif /* CONFIG_IPE_PROP_DM_VERITY */ }; /** diff --git a/security/ipe/ipe.h b/security/ipe/ipe.h index 86c2b8394d93..64626702b689 100644 --- a/security/ipe/ipe.h +++ b/security/ipe/ipe.h @@ -16,4 +16,8 @@ struct ipe_superblock *ipe_sb(const struct super_block *sb); extern bool ipe_enabled; +#ifdef CONFIG_IPE_PROP_DM_VERITY +struct ipe_bdev *ipe_bdev(struct block_device *b); +#endif /* CONFIG_IPE_PROP_DM_VERITY */ + #endif /* _IPE_H */ diff --git a/security/ipe/policy.h b/security/ipe/policy.h index dba5e71e089e..35629e9699f7 100644 --- a/security/ipe/policy.h +++ b/security/ipe/policy.h @@ -33,6 +33,9 @@ enum ipe_action_type { enum ipe_prop_type { IPE_PROP_BOOT_VERIFIED_FALSE, IPE_PROP_BOOT_VERIFIED_TRUE, + IPE_PROP_DMV_ROOTHASH, + IPE_PROP_DMV_SIG_FALSE, + IPE_PROP_DMV_SIG_TRUE, __IPE_PROP_MAX }; diff --git a/security/ipe/policy_parser.c b/security/ipe/policy_parser.c index 3fc8fea55b95..802e31f14b22 100644 --- a/security/ipe/policy_parser.c +++ b/security/ipe/policy_parser.c @@ -11,6 +11,7 @@ #include "policy.h" #include "policy_parser.h" +#include "digest.h" #define START_COMMENT '#' #define IPE_POLICY_DELIM " \t" @@ -216,6 +217,7 @@ static void free_rule(struct ipe_rule *r) list_for_each_entry_safe(p, t, &r->props, next) { list_del(&p->next); + ipe_digest_free(p->value); kfree(p); } @@ -268,6 +270,9 @@ static enum ipe_action_type parse_action(char *t) static const match_table_t property_tokens = { {IPE_PROP_BOOT_VERIFIED_FALSE, "boot_verified=FALSE"}, {IPE_PROP_BOOT_VERIFIED_TRUE, "boot_verified=TRUE"}, + {IPE_PROP_DMV_ROOTHASH, "dmverity_roothash=%s"}, + {IPE_PROP_DMV_SIG_FALSE, "dmverity_signature=FALSE"}, + {IPE_PROP_DMV_SIG_TRUE, "dmverity_signature=TRUE"}, {IPE_PROP_INVALID, NULL} }; @@ -287,6 +292,7 @@ static int parse_property(char *t, struct ipe_rule *r) struct ipe_prop *p = NULL; int rc = 0; int token; + char *dup = NULL; p = kzalloc(sizeof(*p), GFP_KERNEL); if (!p) @@ -295,8 +301,22 @@ static int parse_property(char *t, struct ipe_rule *r) token = match_token(t, property_tokens, args); switch (token) { + case IPE_PROP_DMV_ROOTHASH: + dup = match_strdup(&args[0]); + if (!dup) { + rc = -ENOMEM; + goto err; + } + p->value = ipe_digest_parse(dup); + if (IS_ERR(p->value)) { + rc = PTR_ERR(p->value); + goto err; + } + fallthrough; case IPE_PROP_BOOT_VERIFIED_FALSE: case IPE_PROP_BOOT_VERIFIED_TRUE: + case IPE_PROP_DMV_SIG_FALSE: + case IPE_PROP_DMV_SIG_TRUE: p->type = token; break; default: @@ -307,10 +327,12 @@ static int parse_property(char *t, struct ipe_rule *r) goto err; list_add_tail(&p->next, &r->props); +out: + kfree(dup); return rc; err: kfree(p); - return rc; + goto out; } /** -- 2.44.0
This patch introduces a new hook to save inode's integrity data. For example, for fsverity enabled files, LSMs can use this hook to save the verified fsverity builtin signature into the inode's security blob, and LSMs can make access decisions based on the data inside the signature, like the signer certificate. Signed-off-by: Fan Wu <wufan@linux.microsoft.com> -- v1-v14: + Not present v15: + Introduced --- include/linux/lsm_hook_defs.h | 3 +++ include/linux/security.h | 10 ++++++++++ security/security.c | 28 ++++++++++++++++++++++++++++ 3 files changed, 41 insertions(+) diff --git a/include/linux/lsm_hook_defs.h b/include/linux/lsm_hook_defs.h index 6808ae763913..c88587fc3691 100644 --- a/include/linux/lsm_hook_defs.h +++ b/include/linux/lsm_hook_defs.h @@ -177,6 +177,9 @@ LSM_HOOK(int, 0, inode_listsecurity, struct inode *inode, char *buffer, LSM_HOOK(void, LSM_RET_VOID, inode_getsecid, struct inode *inode, u32 *secid) LSM_HOOK(int, 0, inode_copy_up, struct dentry *src, struct cred **new) LSM_HOOK(int, -EOPNOTSUPP, inode_copy_up_xattr, const char *name) +LSM_HOOK(int, 0, inode_setintegrity, struct inode *inode, + enum lsm_intgr_type type, const void *value, size_t size) + LSM_HOOK(int, 0, kernfs_init_security, struct kernfs_node *kn_dir, struct kernfs_node *kn) LSM_HOOK(int, 0, file_permission, struct file *file, int mask) diff --git a/include/linux/security.h b/include/linux/security.h index 60b40b523d57..0885866b261e 100644 --- a/include/linux/security.h +++ b/include/linux/security.h @@ -405,6 +405,9 @@ int security_inode_listsecurity(struct inode *inode, char *buffer, size_t buffer void security_inode_getsecid(struct inode *inode, u32 *secid); int security_inode_copy_up(struct dentry *src, struct cred **new); int security_inode_copy_up_xattr(const char *name); +int security_inode_setintegrity(struct inode *inode, + enum lsm_intgr_type type, const void *value, + size_t size); int security_kernfs_init_security(struct kernfs_node *kn_dir, struct kernfs_node *kn); int security_file_permission(struct file *file, int mask); @@ -1021,6 +1024,13 @@ static inline int security_inode_copy_up(struct dentry *src, struct cred **new) return 0; } +static inline int security_inode_setintegrity(struct inode *inode, + enum lsm_intgr_type, type, + const void *value, size_t size) +{ + return 0; +} + static inline int security_kernfs_init_security(struct kernfs_node *kn_dir, struct kernfs_node *kn) { diff --git a/security/security.c b/security/security.c index 8d88529ac904..c5d426d084ab 100644 --- a/security/security.c +++ b/security/security.c @@ -2681,6 +2681,34 @@ int security_inode_copy_up_xattr(const char *name) } EXPORT_SYMBOL(security_inode_copy_up_xattr); +/** + * security_inode_setintegrity() - Set the inode's integrity data + * @inode: inode + * @type: type of integrity, e.g. hash digest, signature, etc + * @value: the integrity value + * @size: size of the integrity value + * + * Register a verified integrity measurement of a inode with the LSM. + * + * Return: Returns 0 on success, negative values on failure. + */ +int security_inode_setintegrity(struct inode *inode, + enum lsm_intgr_type type, const void *value, + size_t size) +{ + int rc = 0; + struct security_hook_list *p; + + hlist_for_each_entry(p, &security_hook_heads.inode_setintegrity, list) { + rc = p->hook.inode_setintegrity(inode, type, value, size); + if (rc) + return rc; + } + + return LSM_RET_DEFAULT(inode_setintegrity); +} +EXPORT_SYMBOL(security_inode_setintegrity); + /** * security_kernfs_init_security() - Init LSM context for a kernfs node * @kn_dir: parent kernfs node -- 2.44.0