platform/chrome: cros_ec_proto: Lock device when updating MKBP version

commit df615907f1bf907260af01ccb904d0e9304b5278 upstream.

The cros_ec_get_host_command_version_mask() function requires that t

platform/chrome: cros_ec_proto: Lock device when updating MKBP version

commit df615907f1bf907260af01ccb904d0e9304b5278 upstream.

The cros_ec_get_host_command_version_mask() function requires that the
caller must have ec_dev->lock mutex before calling it. This requirement
was not met and as a result it was possible that two commands were sent
to the device at the same time.

The problem was observed while using UART backend which doesn't use any
additional locks, unlike SPI backend which locks the controller until
response is received.

Fixes: f74c7557ed0d ("platform/chrome: cros_ec_proto: Update version on GET_NEXT_EVENT failure")
Cc: stable@vger.kernel.org
Signed-off-by: Patryk Duda <patrykd@google.com>
Link: https://lore.kernel.org/r/20240730104425.607083-1-patrykd@google.com
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

show more ...

platform/chrome: cros_ec_proto: Update version on GET_NEXT_EVENT failure

Some EC based devices (e.g. Fingerpint MCU) can jump to RO part of the
firmware (intentionally or due to device reboot). The

platform/chrome: cros_ec_proto: Update version on GET_NEXT_EVENT failure

Some EC based devices (e.g. Fingerpint MCU) can jump to RO part of the
firmware (intentionally or due to device reboot). The RO part doesn't
change during the device lifecycle, so it won't support newer version
of EC_CMD_GET_NEXT_EVENT command.

Function cros_ec_query_all() is responsible for finding maximum
supported MKBP event version. It's usually called when the device is
running RW part of the firmware, so the command version can be
potentially higher than version supported by the RO.

The problem was fixed by updating maximum supported version when the
device returns EC_RES_INVALID_VERSION (mapped to -ENOPROTOOPT). That way
the kernel will use highest common version supported by RO and RW.

Fixes: 3300fdd630d4 ("platform/chrome: cros_ec: handle MKBP more events flag")
Cc: <stable@vger.kernel.org> # 5.10+
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Patryk Duda <pdk@semihalf.com>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220802154128.21175-1-pdk@semihalf.com

show more ...

platform/chrome: cros_ec_proto: return -EPROTO if empty payload

cros_ec_wait_until_complete() sends EC_CMD_GET_COMMS_STATUS which expects
to receive sizeof(struct ec_response_get_comms_status) from

platform/chrome: cros_ec_proto: return -EPROTO if empty payload

cros_ec_wait_until_complete() sends EC_CMD_GET_COMMS_STATUS which expects
to receive sizeof(struct ec_response_get_comms_status) from
cros_ec_xfer_command().

Return -EPROTO if cros_ec_xfer_command() returns 0.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220718050914.2267370-11-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: return -EAGAIN when retries timed out

While EC_COMMS_STATUS_PROCESSING flag is still on after it tries
EC_COMMAND_RETRIES times for sending EC_CMD_GET_COMMS_STATUS,
c

platform/chrome: cros_ec_proto: return -EAGAIN when retries timed out

While EC_COMMS_STATUS_PROCESSING flag is still on after it tries
EC_COMMAND_RETRIES times for sending EC_CMD_GET_COMMS_STATUS,
cros_ec_wait_until_complete() doesn't return an error code.

Return -EAGAIN in the case instead.

Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Link: https://lore.kernel.org/r/20220718050914.2267370-9-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: separate cros_ec_wait_until_complete()

EC returns EC_RES_IN_PROGRESS if the host command needs more time to
complete. Whenever receives the return code, cros_ec_send

platform/chrome: cros_ec_proto: separate cros_ec_wait_until_complete()

EC returns EC_RES_IN_PROGRESS if the host command needs more time to
complete. Whenever receives the return code, cros_ec_send_command()
sends EC_CMD_GET_COMMS_STATUS to query the command status.

Separate cros_ec_wait_until_complete() from cros_ec_send_command().
It sends EC_CMD_GET_COMMS_STATUS and waits until the previous command
was completed, or encountered error, or timed out.

Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Link: https://lore.kernel.org/r/20220718050914.2267370-7-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: separate cros_ec_xfer_command()

cros_ec_send_command() has extra logic to handle EC_RES_IN_PROGRESS.
Separate the command transfer part into cros_ec_xfer_command() so

platform/chrome: cros_ec_proto: separate cros_ec_xfer_command()

cros_ec_send_command() has extra logic to handle EC_RES_IN_PROGRESS.
Separate the command transfer part into cros_ec_xfer_command() so
that other functions can re-use it.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220718050914.2267370-6-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: add "cros_ec_" prefix to send_command()

To be neat, add "cros_ec_" prefix to static function send_command().

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-

platform/chrome: cros_ec_proto: add "cros_ec_" prefix to send_command()

To be neat, add "cros_ec_" prefix to static function send_command().

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220718050914.2267370-3-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: Fix spelling mistake "unknwon" -> "unknown"

There is a spelling mistake in a dev_dbg message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-o

platform/chrome: cros_ec_proto: Fix spelling mistake "unknwon" -> "unknown"

There is a spelling mistake in a dev_dbg message. Fix it.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220614064909.47804-1-colin.i.king@gmail.com

show more ...

platform/chrome: cros_ec_proto: handle empty payload in getting wake mask

cros_ec_get_host_event_wake_mask() expects to receive
sizeof(struct ec_response_host_event_mask) from send_command(). The
p

platform/chrome: cros_ec_proto: handle empty payload in getting wake mask

cros_ec_get_host_event_wake_mask() expects to receive
sizeof(struct ec_response_host_event_mask) from send_command(). The
payload is valid only if the return value is positive.

Return -EPROTO if send_command() returns 0 in
cros_ec_get_host_event_wake_mask().

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-22-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: return 0 on getting wake mask success

cros_ec_get_host_event_wake_mask() used to return value from
send_command() which is number of bytes for input payload on succes

platform/chrome: cros_ec_proto: return 0 on getting wake mask success

cros_ec_get_host_event_wake_mask() used to return value from
send_command() which is number of bytes for input payload on success
(i.e. sizeof(struct ec_response_host_event_mask)).

However, the callers don't need to know how many bytes are available.

Don't return number of available bytes. Instead, return 0 on success;
otherwise, negative integers on error.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-20-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: handle empty payload in getting cmd mask

cros_ec_get_host_command_version_mask() expects to receive
sizeof(struct ec_response_get_cmd_versions) from send_command().

platform/chrome: cros_ec_proto: handle empty payload in getting cmd mask

cros_ec_get_host_command_version_mask() expects to receive
sizeof(struct ec_response_get_cmd_versions) from send_command(). The
payload is valid only if the return value is positive.

Return -EPROTO if send_command() returns 0 in
cros_ec_get_host_command_version_mask().

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-19-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: check `msg->result` in getting cmd mask

cros_ec_get_host_command_version_mask() should check if EC wasn't happy
by checking `msg->result`.

Use cros_ec_map_error() an

platform/chrome: cros_ec_proto: check `msg->result` in getting cmd mask

cros_ec_get_host_command_version_mask() should check if EC wasn't happy
by checking `msg->result`.

Use cros_ec_map_error() and return the error code if any.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-17-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: return 0 on getting cmd mask success

cros_ec_get_host_command_version_mask() used to return value from
send_command() which is number of available bytes for input pay

platform/chrome: cros_ec_proto: return 0 on getting cmd mask success

cros_ec_get_host_command_version_mask() used to return value from
send_command() which is number of available bytes for input payload on
success (i.e. sizeof(struct ec_response_get_cmd_versions)).

However, the callers don't need to know how many bytes are available.

Don't return number of available bytes. Instead, return 0 on success;
otherwise, negative integers on error.

Also remove the unneeded `ver_mask` initialization as the callers should
take it only if cros_ec_get_host_command_version_mask() returns 0.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-15-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: don't show MKBP version if unsupported

It wrongly showed the following message when it doesn't support MKBP:
"MKBP support version 4294967295".

Fix it.

Reviewed-by:

platform/chrome: cros_ec_proto: don't show MKBP version if unsupported

It wrongly showed the following message when it doesn't support MKBP:
"MKBP support version 4294967295".

Fix it.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-14-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: handle empty payload in getting info legacy

cros_ec_get_proto_info_legacy() expects to receive
sizeof(struct ec_response_hello) from send_command(). The payload is
v

platform/chrome: cros_ec_proto: handle empty payload in getting info legacy

cros_ec_get_proto_info_legacy() expects to receive
sizeof(struct ec_response_hello) from send_command(). The payload is
valid only if the return value is positive.

Return -EPROTO if send_command() returns 0 in
cros_ec_get_proto_info_legacy().

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-13-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: separate cros_ec_get_proto_info_legacy()

Rename cros_ec_host_command_proto_query_v2() to
cros_ec_get_proto_info_legacy() and make it responsible for setting
`ec_dev`

platform/chrome: cros_ec_proto: separate cros_ec_get_proto_info_legacy()

Rename cros_ec_host_command_proto_query_v2() to
cros_ec_get_proto_info_legacy() and make it responsible for setting
`ec_dev` fields for EC protocol v2.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-11-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: handle empty payload in getting proto info

cros_ec_get_proto_info() expects to receive
sizeof(struct ec_response_get_protocol_info) from send_command(). The
payload

platform/chrome: cros_ec_proto: handle empty payload in getting proto info

cros_ec_get_proto_info() expects to receive
sizeof(struct ec_response_get_protocol_info) from send_command(). The
payload is valid only if the return value is positive.

Return -EPROTO if send_command() returns 0 in cros_ec_get_proto_info().

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-10-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: separate cros_ec_get_proto_info()

Rename cros_ec_host_command_proto_query() to cros_ec_get_proto_info()
and make it responsible for setting `ec_dev` fields according

platform/chrome: cros_ec_proto: separate cros_ec_get_proto_info()

Rename cros_ec_host_command_proto_query() to cros_ec_get_proto_info()
and make it responsible for setting `ec_dev` fields according to the
response protocol info.

Also make cros_ec_get_host_event_wake_mask() allocate its own message
buffer. It was lucky that size of `struct ec_response_host_event_mask`
is less than `struct ec_response_get_protocol_info`. Thus, the buffer
wasn't overflow.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-8-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: use cros_ec_map_error()

Use cros_ec_map_error() in cros_ec_get_host_event_wake_mask().

The behavior of cros_ec_get_host_event_wake_mask() slightly changed. It
is ac

platform/chrome: cros_ec_proto: use cros_ec_map_error()

Use cros_ec_map_error() in cros_ec_get_host_event_wake_mask().

The behavior of cros_ec_get_host_event_wake_mask() slightly changed. It
is acceptable because the caller only needs it returns negative integers
for indicating errors. Especially, the EC_RES_INVALID_COMMAND still
maps to -EOPNOTSUPP.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-7-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: remove redundant NULL check

send_command() already checks if `ec_dev->pkt_xfer` is NULL. Remove the
redundant check.

Reviewed-by: Guenter Roeck <groeck@chromium.org

platform/chrome: cros_ec_proto: remove redundant NULL check

send_command() already checks if `ec_dev->pkt_xfer` is NULL. Remove the
redundant check.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-6-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: assign buffer size from protocol info

`din_size` is calculated from `ec_dev->max_response`.
`ec_dev->max_response` is further calculated from the protocol info.

To m

platform/chrome: cros_ec_proto: assign buffer size from protocol info

`din_size` is calculated from `ec_dev->max_response`.
`ec_dev->max_response` is further calculated from the protocol info.

To make it clear, assign `din_size` and `dout_size` from protocol info
directly.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-5-tzungbi@kernel.org

show more ...

platform/chrome: use macros for passthru indexes

Move passthru indexes for EC and PD devices to common header. Also use
them instead of literal constants.

Reviewed-by: Guenter Roeck <groeck@chromi

platform/chrome: use macros for passthru indexes

Move passthru indexes for EC and PD devices to common header. Also use
them instead of literal constants.

Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220609084957.3684698-4-tzungbi@kernel.org

show more ...

platform/chrome: cros_ec_proto: Update size arg types

cros_ec_cmd() takes 2 size arguments. Update them to be of the more
appropriate type size_t.

Suggested-by: Stephen Boyd <swboyd@chromium.org>
S

platform/chrome: cros_ec_proto: Update size arg types

cros_ec_cmd() takes 2 size arguments. Update them to be of the more
appropriate type size_t.

Suggested-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Prashant Malani <pmalani@chromium.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220606201825.763788-4-pmalani@chromium.org

show more ...

platform/chrome: cros_ec_proto: Rename cros_ec_command function

cros_ec_command() is the name of a function as well as a struct, as such
it can confuse indexing tools (like ctags). Avoid this by ren

platform/chrome: cros_ec_proto: Rename cros_ec_command function

cros_ec_command() is the name of a function as well as a struct, as such
it can confuse indexing tools (like ctags). Avoid this by renaming it to
cros_ec_cmd(). Update all the callsites to use the new name.

This patch is a find-and-replace, so should not introduce any functional
changes.

Suggested-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Prashant Malani <pmalani@chromium.org>
Acked-by: Lee Jones <lee.jones@linaro.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220606201825.763788-3-pmalani@chromium.org

show more ...

platform/chrome: cros_ec_proto: update cros_ec_check_result() comment

At first glance, cros_ec_check_result() is quite like cros_ec_map_error().
They check for `ec_msg->result` and return correspond

platform/chrome: cros_ec_proto: update cros_ec_check_result() comment

At first glance, cros_ec_check_result() is quite like cros_ec_map_error().
They check for `ec_msg->result` and return corresponding errors. However,
as calling from `pkt_xfer` and `cmd_xfer`, cros_ec_check_result() should
not report furthermore errors. -EAGAIN is the only exception.

See [1][2][3] for some known userland programs' code. The return code
from ioctl only denotes the EC communication status. Userland programs
would further analyze the `result` in struct cros_ec_command* for
follow-up actions (e.g. [4]).

To clarify, update the function comment.

[1]: https://crrev.com/54400e93a75ef440a83d6eaac2cec066daf99cf0/util/comm-dev.c#154
[2]: https://crrev.com/fe32670a89bf59e1aff84bba9dd3295657b85e9b/cros_ec_dev.c#296
[3]: https://crrev.com/4e19eb1d89de0422ff1bbd3f7260b131c761098c/drivers/google/cros_ec_dev.c#120
[4]: https://crrev.com/54400e93a75ef440a83d6eaac2cec066daf99cf0/util/comm-dev.c#164

Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Link: https://lore.kernel.org/r/20220518091814.2028579-4-tzungbi@kernel.org

show more ...