Skip to content

AP bus/zcrypt/pkey/paes no-mem-alloc patches

Harald Freudenberger linux-s390 8 versions – Applied

Status from this message: Applied, thanks!

Get series with b4
b4 am [email protected]

Versions newest first · linked by author and title

  1. v8 Latest25 patches 1 replyApplied
  2. v725 patches 5 repliesSuperseded
  3. v625 patches 5 repliesSuperseded
Earlier versions · 5
VersionPostedPatchesRepliesStatus
v525 patches9 repliesSuperseded
v424 patches25 repliesSuperseded
v321 patches9 repliesSuperseded
v220 patches34 repliesSuperseded
v120 patches10 repliesSuperseded

Changelog from the cover letter · 37 lines

v1: initial version
v2: - Rework on patch 0001 and 0002 based on feedback from Holger.
Also there was one place in zcrypt_msgtype50.c where still an ap msg buffer was alloacated.
- Rework on patch 0003 - fixed feedback from Holger. Also the min poolitems is now a module parameter and defaults to 8.
- Rework on patch 0004 - as suggested by Holger the "userspace" parameter is now included into the ap msg flags.
- Rework on patch 0005 - nr of cca cprbs in the mempool is now a module parameter.
- Rework on patch 0006 - nr of ep11 cprbs in the mempool is now a module parameter.
- Rework on patch 0007 - as suggested by Holger instead of implementing a copy-and-pasted new function zcrypt_device_status_mask_ext2() use and extend the existing the existing function to avoid code duplication.
- The rest of the patch series needed adaptions but there is no functional change compared to v1.
v3: Rework based on feedback from Holger:
- There is now one zcrypt module parameter "mempool_threshold" controlling the cprb mempools for CCA and EP11. The default value of 5 shall cover 5 CCA or EP11 requests/replies in parallel.
- The cca and ep11 card and domain info cache is done - anyway nearly all callers used the verify=1 to enforce the retrieval of fresh information. So now this info is always freshly fetched from the card/domain.
- Holger found out that there are some unused functions lurking around - deleted :-)
- The only thing still missing is no-mem support for the UV driver. As of now the pkey uv layer refuses to contact the UV dd in case the NOMEMALLOC flag is given.
- Tested and found no issues.
v4: Still some minor changes:
- The UV patch is now part of this patch series
- With that came a slight rework of the pkey uv handler code. It is now pre-allocating one struct for use with the UV driver.
- The zcrypt api module parameter mempool_threshold is now checked and enforced to have at least a value of 1.
- Heiko found some things, for example the mempool_create() was for some reason checking for PTR_ERR which does not apply to the used way of creating the mempools - fixed in all patches.
- Heiko also had the question about the artificial limit of 16 APQNs as target addressing for EP11 cprbs. Together with Holger we found out, that this is not needed at all - so yet another patch now reworks this code (see patch "s390/zcrypt:
Avoid alloc and copy of ep11 targets if kernelspace cprb").
v5: - new patch for the uv code to support another non-malloc
interface function ("s390/uv: Rename find_secret() to uv_find_secret() and publish")
- with that a slight rework of the pkey uv handler which is now able to retrieve uv protected key secrets without malloc.
- new patch for the uv code to remove the old interface function ("s390/uv: Remove uv_get_secret_metadata function").
- Lots of minor reworks triggered by review comments from Holger.
v6: - Typo in patch header and other patch header text improvements
fixed
- zero page to be used for uv before calling uv_find_secret().
- added the latest signed-offs
- tested, looks good
v7: - Requested changes from Holger on patch #4 where missing.
Added. No other changes.
v8: - Fixed warnings from clang compiler about uninit variable,
fixed findings from Heiko about possible memset on NULL ptr.
Changed: patch 0002, 0006, 0007

Patches in v8 Applied

#Subject and filesSizeRepliesApplied
1/25s390/ap: Move response_type struct into ap_msg structdrivers/s390/crypto/ap_bus.h drivers/s390/crypto/zcrypt_msgtype50.c drivers/s390/crypto/zcrypt_msgtype6.c+37 -610 replies
2/25s390/ap/zcrypt: Rework AP message buffer allocationdrivers/s390/crypto/ap_bus.c drivers/s390/crypto/ap_bus.h drivers/s390/crypto/zcrypt_api.c +2 more+87 -660 replies
3/25s390/ap: Introduce ap message buffer pooldrivers/s390/crypto/ap_bus.c drivers/s390/crypto/ap_bus.h drivers/s390/crypto/zcrypt_api.c+58 -90 replies
4/25s390/zcrypt: Avoid alloc and copy of ep11 targets if kernelspace cprbdrivers/s390/crypto/zcrypt_api.c+19 -230 replies
5/25s390/ap/zcrypt: New xflag parameterdrivers/s390/crypto/ap_bus.c drivers/s390/crypto/ap_bus.h drivers/s390/crypto/zcrypt_api.c +3 more+52 -380 replies
6/25s390/zcrypt: Introduce cprb mempool for cca misc functionsdrivers/s390/crypto/zcrypt_api.c drivers/s390/crypto/zcrypt_api.h drivers/s390/crypto/zcrypt_ccamisc.c +1 more+103 -320 replies
7/25s390/zcrypt: Introduce cprb mempool for ep11 misc functionsdrivers/s390/crypto/zcrypt_api.c drivers/s390/crypto/zcrypt_ep11misc.c drivers/s390/crypto/zcrypt_ep11misc.h+96 -620 replies
8/25s390/zcrypt: Rework zcrypt function zcrypt_device_status_mask_extdrivers/s390/crypto/zcrypt_api.c drivers/s390/crypto/zcrypt_api.h drivers/s390/crypto/zcrypt_ccamisc.c +1 more+33 -130 replies
9/25s390/zcrypt: Introduce pre-allocated device status array for cca miscdrivers/s390/crypto/zcrypt_ccamisc.c+37 -100 replies
10/25s390/zcrypt: Introduce pre-allocated device status array for ep11 miscdrivers/s390/crypto/zcrypt_ep11misc.c+36 -100 replies
11/25s390/zcrypt: Remove unused functions from cca miscdrivers/s390/crypto/zcrypt_ccamisc.c drivers/s390/crypto/zcrypt_ccamisc.h+0 -1510 replies
12/25s390/zcrypt: Remove CCA and EP11 card and domain info cachesdrivers/s390/crypto/pkey_cca.c drivers/s390/crypto/zcrypt_ccamisc.c drivers/s390/crypto/zcrypt_ccamisc.h +3 more+47 -2320 replies
13/25s390/zcrypt: Rework cca findcard() implementation and callersdrivers/s390/crypto/pkey_cca.c drivers/s390/crypto/zcrypt_ccamisc.c drivers/s390/crypto/zcrypt_ccamisc.h+23 -450 replies
14/25s390/zcrypt: Rework ep11 findcard() implementation and callersdrivers/s390/crypto/pkey_ep11.c drivers/s390/crypto/zcrypt_ep11misc.c drivers/s390/crypto/zcrypt_ep11misc.h+20 -420 replies
15/25s390/zcrypt: Rework cca misc functions kmallocs to use the cprb mempooldrivers/s390/crypto/zcrypt_ccamisc.c+29 -130 replies
16/25s390/zcrypt: Propagate xflags argument with cca_get_info()drivers/s390/crypto/zcrypt_ccamisc.c drivers/s390/crypto/zcrypt_ccamisc.h drivers/s390/crypto/zcrypt_cex4.c+6 -60 replies
17/25s390/zcrypt: Locate ep11_domain_query_info onto the stack instead of kmallocdrivers/s390/crypto/zcrypt_ep11misc.c+10 -150 replies
18/25s390/zcrypt: Rework ep11 misc functions to use cprb mempooldrivers/s390/crypto/pkey_ep11.c drivers/s390/crypto/zcrypt_cex4.c drivers/s390/crypto/zcrypt_ep11misc.c +1 more+76 -490 replies
19/25s390/pkey: Rework CCA pkey handler to use stack for small memory allocsdrivers/s390/crypto/pkey_cca.c+10 -330 replies
20/25s390/pkey: Rework EP11 pkey handler to use stack for small memory allocsdrivers/s390/crypto/pkey_ep11.c+10 -330 replies
21/25s390/uv: Rename find_secret() to uv_find_secret() and publisharch/s390/include/asm/uv.h arch/s390/kernel/uv.c+17 -50 replies
22/25s390/pkey: Use preallocated memory for retrieve of UV secret metadatadrivers/s390/crypto/pkey_uv.c+35 -30 replies
23/25s390/uv: Remove uv_get_secret_metadata functionarch/s390/include/asm/uv.h arch/s390/kernel/uv.c+0 -320 replies
24/25s390/pkey: Provide and pass xflags within pkey and zcrypt layersarch/s390/include/asm/pkey.h drivers/s390/crypto/pkey_api.c drivers/s390/crypto/pkey_base.c +10 more+224 -1510 replies
25/25s390/pkey/crypto: Introduce xflags param for pkey in-kernel APIarch/s390/crypto/paes_s390.c arch/s390/include/asm/pkey.h drivers/s390/crypto/pkey_api.c+8 -60 replies

Reviews trailers per patch

All 25 patches have a Reviewed-by or Acked-by.

PatchHolger DenglerSteffen Eiden
1/25R-b
2/25R-b
3/25R-b
4/25R-b
5/25R-b
6/25R-b
7/25R-b
8/25R-b
9/25R-b
10/25R-b
11/25R-b
12/25R-b
13/25R-b
14/25R-b
15/25R-b
16/25R-b
17/25R-b
18/25R-b
19/25R-b
20/25R-b
21/25A-b R-b
22/25R-b R-b
23/25A-b R-b
24/25R-b
25/25R-b

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v8 thread

Cover letter [PATCH v8 00/25] AP bus/zcrypt/pkey/paes no-mem-alloc patches

This series of patches has the goal to open up a do-not-allocate memory path from the callers of the pkey in-kernel api down to the crypto cards and back.

The asynch in-kernel cipher implementations (and the s390 PAES cipher implementations are one of them) may be called in a context where memory allocations which trigger IO is not acceptable.

So this patch series reworks the AP bus code, the zcrypt layer, the pkey layer and the pkey handlers to respect this situation by processing a new parameter xflags (execution hints flags). There is a flag PKEY_XFLAG_NOMEMALLOC which tells the code to not allocate memory which may lead to IO operations.

To reach this goal, the actual code changes have been differed. The zcrypt misc functions which need memory for cprb build use a pre allocated memory pool for this purpose. The findcard() functions have one temp memory area preallocated and protected with a mutex. Some smaller data is not allocated any more but went to the stack instead. The AP bus also uses a pre-allocated memory pool for building AP message requests.

Note that the PAES implementation still needs to get reworked to run the protected key derivation in a real asynchronous way. However, this rework of AP bus, zcrypt and pkey is the base work required before reconsidering the PAES implementation.

The patch series starts bottom (AP bus) and goes up the call chain (PKEY). At any time in the patch stack it should compile. For easier review I tried to have one logic code change by each patch and thus keep the patches "small". For the upstream version I intend to fold them together into only a few commits.

Changelog:
v1: initial version
v2: - Rework on patch 0001 and 0002 based on feedback from Holger.
Also there was one place in zcrypt_msgtype50.c where still an ap msg buffer was alloacated.
- Rework on patch 0003 - fixed feedback from Holger. Also the min poolitems is now a module parameter and defaults to 8.
- Rework on patch 0004 - as suggested by Holger the "userspace" parameter is now included into the ap msg flags.
- Rework on patch 0005 - nr of cca cprbs in the mempool is now a module parameter.
- Rework on patch 0006 - nr of ep11 cprbs in the mempool is now a module parameter.
- Rework on patch 0007 - as suggested by Holger instead of implementing a copy-and-pasted new function zcrypt_device_status_mask_ext2() use and extend the existing the existing function to avoid code duplication.
- The rest of the patch series needed adaptions but there is no functional change compared to v1.
v3: Rework based on feedback from Holger:
- There is now one zcrypt module parameter "mempool_threshold" controlling the cprb mempools for CCA and EP11. The default value of 5 shall cover 5 CCA or EP11 requests/replies in parallel.
- The cca and ep11 card and domain info cache is done - anyway nearly all callers used the verify=1 to enforce the retrieval of fresh information. So now this info is always freshly fetched from the card/domain.
- Holger found out that there are some unused functions lurking around - deleted :-)
- The only thing still missing is no-mem support for the UV driver. As of now the pkey uv layer refuses to contact the UV dd in case the NOMEMALLOC flag is given.
- Tested and found no issues.
v4: Still some minor changes:
- The UV patch is now part of this patch series
- With that came a slight rework of the pkey uv handler code. It is now pre-allocating one struct for use with the UV driver.
- The zcrypt api module parameter mempool_threshold is now checked and enforced to have at least a value of 1.
- Heiko found some things, for example the mempool_create() was for some reason checking for PTR_ERR which does not apply to the used way of creating the mempools - fixed in all patches.
- Heiko also had the question about the artificial limit of 16 APQNs as target addressing for EP11 cprbs. Together with Holger we found out, that this is not needed at all - so yet another patch now reworks this code (see patch "s390/zcrypt:
Avoid alloc and copy of ep11 targets if kernelspace cprb").
v5: - new patch for the uv code to support another non-malloc
interface function ("s390/uv: Rename find_secret() to uv_find_secret() and publish")
- with that a slight rework of the pkey uv handler which is now able to retrieve uv protected key secrets without malloc.
- new patch for the uv code to remove the old interface function ("s390/uv: Remove uv_get_secret_metadata function").
- Lots of minor reworks triggered by review comments from Holger.
v6: - Typo in patch header and other patch header text improvements
fixed
- zero page to be used for uv before calling uv_find_secret().
- added the latest signed-offs
- tested, looks good
v7: - Requested changes from Holger on patch #4 where missing.
Added. No other changes.
v8: - Fixed warnings from clang compiler about uninit variable,
fixed findings from Heiko about possible memset on NULL ptr.
Changed: patch 0002, 0006, 0007

Harald Freudenberger (25):
s390/ap: Move response_type struct into ap_msg struct s390/ap/zcrypt: Rework AP message buffer allocation s390/ap: Introduce ap message buffer pool
s390/zcrypt: Avoid alloc and copy of ep11 targets if kernelspace cprb s390/ap/zcrypt: New xflag parameter
s390/zcrypt: Introduce cprb mempool for cca misc functions s390/zcrypt: Introduce cprb mempool for ep11 misc functions s390/zcrypt: Rework zcrypt function zcrypt_device_status_mask_ext s390/zcrypt: Introduce pre-allocated device status array for cca misc s390/zcrypt: Introduce pre-allocated device status array for ep11 misc s390/zcrypt: Remove unused functions from cca misc s390/zcrypt: Remove CCA and EP11 card and domain info caches s390/zcrypt: Rework cca findcard() implementation and callers s390/zcrypt: Rework ep11 findcard() implementation and callers s390/zcrypt: Rework cca misc functions kmallocs to use the cprb
mempool
s390/zcrypt: Propagate xflags argument with cca_get_info() s390/zcrypt: Locate ep11_domain_query_info onto the stack instead of
kmalloc
s390/zcrypt: Rework ep11 misc functions to use cprb mempool s390/pkey: Rework CCA pkey handler to use stack for small memory
allocs
s390/pkey: Rework EP11 pkey handler to use stack for small memory
allocs
s390/uv: Rename find_secret() to uv_find_secret() and publish s390/pkey: Use preallocated memory for retrieve of UV secret metadata s390/uv: Remove uv_get_secret_metadata function
s390/pkey: Provide and pass xflags within pkey and zcrypt layers s390/pkey/crypto: Introduce xflags param for pkey in-kernel API

 arch/s390/crypto/paes_s390.c           |   6 +-
 arch/s390/include/asm/pkey.h           |  15 +-
 arch/s390/include/asm/uv.h             |   5 +-
 arch/s390/kernel/uv.c                  |  47 +--
 drivers/s390/crypto/ap_bus.c           |  74 ++++
 drivers/s390/crypto/ap_bus.h           |  30 +-
 drivers/s390/crypto/pkey_api.c         |  50 +--
 drivers/s390/crypto/pkey_base.c        |  34 +-
 drivers/s390/crypto/pkey_base.h        |  37 +-
 drivers/s390/crypto/pkey_cca.c         | 134 ++++---
 drivers/s390/crypto/pkey_ep11.c        | 117 +++---
 drivers/s390/crypto/pkey_pckmo.c       |   9 +-
 drivers/s390/crypto/pkey_sysfs.c       |   4 +-
 drivers/s390/crypto/pkey_uv.c          |  44 ++-
 drivers/s390/crypto/zcrypt_api.c       | 167 +++++----
 drivers/s390/crypto/zcrypt_api.h       |  16 +-
 drivers/s390/crypto/zcrypt_ccamisc.c   | 486 +++++++++----------------
 drivers/s390/crypto/zcrypt_ccamisc.h   |  49 +--
 drivers/s390/crypto/zcrypt_cex4.c      |  39 +-
 drivers/s390/crypto/zcrypt_ep11misc.c  | 454 +++++++++++------------
 drivers/s390/crypto/zcrypt_ep11misc.h  |  27 +-
 drivers/s390/crypto/zcrypt_msgtype50.c |  36 +-
 drivers/s390/crypto/zcrypt_msgtype6.c  | 109 +++---
 23 files changed, 961 insertions(+), 1028 deletions(-)

base-commit: bffb31040cb61aa72de8ffac655fe1459d581505 --
2.43.0