linux-toradex.git/fs/pstore/platform.c, branch v5.0

pstore: Convert buf_lock to semaphore

2018-12-04T01:11:02+00:00

Instead of running with interrupts disabled, use a semaphore. This should
make it easier for backends that may need to sleep (e.g. EFI) when
performing a write:

|BUG: sleeping function called from invalid context at kernel/sched/completion.c:99
|in_atomic(): 1, irqs_disabled(): 1, pid: 2236, name: sig-xstate-bum
|Preemption disabled at:
|[] pstore_dump+0x72/0x330
|CPU: 26 PID: 2236 Comm: sig-xstate-bum Tainted: G      D           4.20.0-rc3 #45
|Call Trace:
| dump_stack+0x4f/0x6a
| ___might_sleep.cold.91+0xd3/0xe4
| __might_sleep+0x50/0x90
| wait_for_completion+0x32/0x130
| virt_efi_query_variable_info+0x14e/0x160
| efi_query_variable_store+0x51/0x1a0
| efivar_entry_set_safe+0xa3/0x1b0
| efi_pstore_write+0x109/0x140
| pstore_dump+0x11c/0x330
| kmsg_dump+0xa4/0xd0
| oops_exit+0x22/0x30
...

Reported-by: Sebastian Andrzej Siewior 
Fixes: 21b3ddd39fee ("efi: Don't use spinlocks for efi vars")
Signed-off-by: Kees Cook

pstore: Map PSTORE_TYPE_* to strings

2018-12-04T00:52:35+00:00

In later patches we will need to map types to names, so create a
constant table for that which can also be used in different parts of
old and new code. This saves the type in the PRZ which will be useful
in later patches.

Instead of having an explicit PSTORE_TYPE_UNKNOWN, just use ..._MAX.

This includes removing the now redundant filename templates which can use
a single format string. Also, there's no reason to limit the "is it still
compressed?" test to only PSTORE_TYPE_DMESG when building the pstorefs
filename. Records are zero-initialized, so a backend would need to have
explicitly set compressed=1.

Signed-off-by: Joel Fernandes (Google) 
Co-developed-by: Kees Cook 
Signed-off-by: Kees Cook

pstore: Improve and update some comments and status output

2018-12-04T00:52:35+00:00

This improves and updates some comments:
 - dump handler comment out of sync from calling convention
 - fix kern-doc typo

and improves status output:
 - reminder that only kernel crash dumps are compressed
 - do not be silent about ECC infrastructure failures

Signed-off-by: Kees Cook

pstore: Remove needless lock during console writes

2018-12-04T00:52:35+00:00

Since the console writer does not use the preallocated crash dump buffer
any more, there is no reason to perform locking around it.

Fixes: 70ad35db3321 ("pstore: Convert console write to use ->write_buf")
Signed-off-by: Kees Cook 
Reviewed-by: Joel Fernandes (Google)

pstore: Do not use crash buffer for decompression

2018-12-04T00:52:35+00:00

The pre-allocated compression buffer used for crash dumping was also
being used for decompression. This isn't technically safe, since it's
possible the kernel may attempt a crashdump while pstore is populating the
pstore filesystem (and performing decompression). Instead, just allocate
a separate buffer for decompression. Correctness is preferred over
performance here.

Signed-off-by: Kees Cook

pstore: Refactor compression initialization

2018-10-22T14:11:58+00:00

This refactors compression initialization slightly to better handle
getting potentially called twice (via early pstore_register() calls
and later pstore_init()) and improves the comments and reporting to be
more verbose.

Signed-off-by: Kees Cook 
Tested-by: Guenter Roeck

pstore: Allocate compression during late_initcall()

2018-10-22T14:11:58+00:00

ramoops's call of pstore_register() was recently moved to run during
late_initcall() because the crypto backend may not have been ready during
postcore_initcall(). This meant early-boot crash dumps were not getting
caught by pstore any more.

Instead, lets allow calls to pstore_register() earlier, and once crypto
is ready we can initialize the compression.

Reported-by: Sai Prakash Ranjan 
Signed-off-by: Joel Fernandes (Google) 
Tested-by: Sai Prakash Ranjan 
Fixes: cb3bee0369bc ("pstore: Use crypto compress API")
[kees: trivial rebase]
Signed-off-by: Kees Cook 
Tested-by: Guenter Roeck

pstore: Centralize init/exit routines

2018-10-22T14:11:58+00:00

In preparation for having additional actions during init/exit, this moves
the init/exit into platform.c, centralizing the logic to make call outs
to the fs init/exit.

Signed-off-by: Kees Cook 
Tested-by: Guenter Roeck

pstore: add zstd compression support

2018-08-04T01:12:18+00:00

This patch added the 6th compression algorithm support for pstore: zstd.

Signed-off-by: Geliang Tang 
Signed-off-by: Kees Cook

pstore: Convert internal records to timespec64

2018-06-05T23:57:31+00:00

This prepares pstore for converting the VFS layer to timespec64.

Signed-off-by: Kees Cook 
Signed-off-by: Deepa Dinamani