* [PATCH 0/2] meson-sm: Fix issues when reading data from sm
@ 2017-03-03 15:17 Carlo Caione
2017-03-03 15:17 ` [PATCH 1/2] firmware: meson-sm: Check for buffer output size Carlo Caione
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Carlo Caione @ 2017-03-03 15:17 UTC (permalink / raw)
To: linux-arm-kernel
From: Carlo Caione <carlo@endlessm.com>
While working on a related driver I noticed a couple of problems in the
meson-sm driver. The problems are unrelated but since the fixes touch the same
codebase I decided to submit these in the same patchset.
Carlo Caione (2):
firmware: meson-sm: Check for buffer output size
firmware: meson-sm: Allow 0 as valid return value
drivers/firmware/meson/meson_sm.c | 20 ++++++++++++++++----
drivers/nvmem/meson-efuse.c | 2 +-
include/linux/firmware/meson/meson_sm.h | 4 ++--
3 files changed, 19 insertions(+), 7 deletions(-)
--
2.12.0
^ permalink raw reply [flat|nested] 6+ messages in thread* [PATCH 1/2] firmware: meson-sm: Check for buffer output size 2017-03-03 15:17 [PATCH 0/2] meson-sm: Fix issues when reading data from sm Carlo Caione @ 2017-03-03 15:17 ` Carlo Caione 2017-03-22 16:28 ` Srinivas Kandagatla 2017-03-03 15:17 ` [PATCH 2/2] firmware: meson-sm: Allow 0 as valid return value Carlo Caione 2017-03-22 15:55 ` [PATCH 0/2] meson-sm: Fix issues when reading data from sm Kevin Hilman 2 siblings, 1 reply; 6+ messages in thread From: Carlo Caione @ 2017-03-03 15:17 UTC (permalink / raw) To: linux-arm-kernel From: Carlo Caione <carlo@endlessm.com> After the data is read by the secure monitor driver it is being copied in the output buffer checking only the size of the bounce buffer but not the size of the output buffer. Fix this in the secure monitor driver slightly changing the API. Fix also the efuse driver that it is the only driver using this API to not break bisectability. Signed-off-by: Carlo Caione <carlo@endlessm.com> --- drivers/firmware/meson/meson_sm.c | 10 +++++++--- drivers/nvmem/meson-efuse.c | 2 +- include/linux/firmware/meson/meson_sm.h | 4 ++-- 3 files changed, 10 insertions(+), 6 deletions(-) diff --git a/drivers/firmware/meson/meson_sm.c b/drivers/firmware/meson/meson_sm.c index b0d254930ed3..5f30a5774e57 100644 --- a/drivers/firmware/meson/meson_sm.c +++ b/drivers/firmware/meson/meson_sm.c @@ -127,6 +127,7 @@ EXPORT_SYMBOL(meson_sm_call); * meson_sm_call_read - retrieve data from secure-monitor * * @buffer: Buffer to store the retrieved data + * @bsize: Size of the buffer * @cmd_index: Index of the SMC32 function ID * @arg0: SMC32 Argument 0 * @arg1: SMC32 Argument 1 @@ -136,8 +137,8 @@ EXPORT_SYMBOL(meson_sm_call); * * Return: size of read data on success, a negative value on error */ -int meson_sm_call_read(void *buffer, unsigned int cmd_index, u32 arg0, - u32 arg1, u32 arg2, u32 arg3, u32 arg4) +int meson_sm_call_read(void *buffer, unsigned int bsize, unsigned int cmd_index, + u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4) { u32 size; @@ -147,10 +148,13 @@ int meson_sm_call_read(void *buffer, unsigned int cmd_index, u32 arg0, if (!fw.chip->cmd_shmem_out_base) return -EINVAL; + if (bsize > fw.chip->shmem_size) + return -EINVAL; + if (meson_sm_call(cmd_index, &size, arg0, arg1, arg2, arg3, arg4) < 0) return -EINVAL; - if (!size || size > fw.chip->shmem_size) + if (!size || size > bsize) return -EINVAL; if (buffer) diff --git a/drivers/nvmem/meson-efuse.c b/drivers/nvmem/meson-efuse.c index f207c3b10482..70bfc9839bb2 100644 --- a/drivers/nvmem/meson-efuse.c +++ b/drivers/nvmem/meson-efuse.c @@ -27,7 +27,7 @@ static int meson_efuse_read(void *context, unsigned int offset, u8 *buf = val; int ret; - ret = meson_sm_call_read(buf, SM_EFUSE_READ, offset, + ret = meson_sm_call_read(buf, bytes, SM_EFUSE_READ, offset, bytes, 0, 0, 0); if (ret < 0) return ret; diff --git a/include/linux/firmware/meson/meson_sm.h b/include/linux/firmware/meson/meson_sm.h index 8e953c6f394a..37a5eaea69dd 100644 --- a/include/linux/firmware/meson/meson_sm.h +++ b/include/linux/firmware/meson/meson_sm.h @@ -25,7 +25,7 @@ int meson_sm_call(unsigned int cmd_index, u32 *ret, u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4); int meson_sm_call_write(void *buffer, unsigned int b_size, unsigned int cmd_index, u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4); -int meson_sm_call_read(void *buffer, unsigned int cmd_index, u32 arg0, u32 arg1, - u32 arg2, u32 arg3, u32 arg4); +int meson_sm_call_read(void *buffer, unsigned int bsize, unsigned int cmd_index, + u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4); #endif /* _MESON_SM_FW_H_ */ -- 2.12.0 ^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH 1/2] firmware: meson-sm: Check for buffer output size 2017-03-03 15:17 ` [PATCH 1/2] firmware: meson-sm: Check for buffer output size Carlo Caione @ 2017-03-22 16:28 ` Srinivas Kandagatla 0 siblings, 0 replies; 6+ messages in thread From: Srinivas Kandagatla @ 2017-03-22 16:28 UTC (permalink / raw) To: linux-arm-kernel On 03/03/17 15:17, Carlo Caione wrote: > From: Carlo Caione <carlo@endlessm.com> > > After the data is read by the secure monitor driver it is being copied > in the output buffer checking only the size of the bounce buffer but not > the size of the output buffer. > > Fix this in the secure monitor driver slightly changing the API. Fix > also the efuse driver that it is the only driver using this API to not > break bisectability. > > Signed-off-by: Carlo Caione <carlo@endlessm.com> Sorry for the delay!! For nvmem part, Acked-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> > --- > drivers/firmware/meson/meson_sm.c | 10 +++++++--- > drivers/nvmem/meson-efuse.c | 2 +- > include/linux/firmware/meson/meson_sm.h | 4 ++-- > 3 files changed, 10 insertions(+), 6 deletions(-) > > diff --git a/drivers/firmware/meson/meson_sm.c b/drivers/firmware/meson/meson_sm.c > index b0d254930ed3..5f30a5774e57 100644 > --- a/drivers/firmware/meson/meson_sm.c > +++ b/drivers/firmware/meson/meson_sm.c > @@ -127,6 +127,7 @@ EXPORT_SYMBOL(meson_sm_call); > * meson_sm_call_read - retrieve data from secure-monitor > * > * @buffer: Buffer to store the retrieved data > + * @bsize: Size of the buffer > * @cmd_index: Index of the SMC32 function ID > * @arg0: SMC32 Argument 0 > * @arg1: SMC32 Argument 1 > @@ -136,8 +137,8 @@ EXPORT_SYMBOL(meson_sm_call); > * > * Return: size of read data on success, a negative value on error > */ > -int meson_sm_call_read(void *buffer, unsigned int cmd_index, u32 arg0, > - u32 arg1, u32 arg2, u32 arg3, u32 arg4) > +int meson_sm_call_read(void *buffer, unsigned int bsize, unsigned int cmd_index, > + u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4) > { > u32 size; > > @@ -147,10 +148,13 @@ int meson_sm_call_read(void *buffer, unsigned int cmd_index, u32 arg0, > if (!fw.chip->cmd_shmem_out_base) > return -EINVAL; > > + if (bsize > fw.chip->shmem_size) > + return -EINVAL; > + > if (meson_sm_call(cmd_index, &size, arg0, arg1, arg2, arg3, arg4) < 0) > return -EINVAL; > > - if (!size || size > fw.chip->shmem_size) > + if (!size || size > bsize) > return -EINVAL; > > if (buffer) > diff --git a/drivers/nvmem/meson-efuse.c b/drivers/nvmem/meson-efuse.c > index f207c3b10482..70bfc9839bb2 100644 > --- a/drivers/nvmem/meson-efuse.c > +++ b/drivers/nvmem/meson-efuse.c > @@ -27,7 +27,7 @@ static int meson_efuse_read(void *context, unsigned int offset, > u8 *buf = val; > int ret; > > - ret = meson_sm_call_read(buf, SM_EFUSE_READ, offset, > + ret = meson_sm_call_read(buf, bytes, SM_EFUSE_READ, offset, > bytes, 0, 0, 0); > if (ret < 0) > return ret; > diff --git a/include/linux/firmware/meson/meson_sm.h b/include/linux/firmware/meson/meson_sm.h > index 8e953c6f394a..37a5eaea69dd 100644 > --- a/include/linux/firmware/meson/meson_sm.h > +++ b/include/linux/firmware/meson/meson_sm.h > @@ -25,7 +25,7 @@ int meson_sm_call(unsigned int cmd_index, u32 *ret, u32 arg0, u32 arg1, > u32 arg2, u32 arg3, u32 arg4); > int meson_sm_call_write(void *buffer, unsigned int b_size, unsigned int cmd_index, > u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4); > -int meson_sm_call_read(void *buffer, unsigned int cmd_index, u32 arg0, u32 arg1, > - u32 arg2, u32 arg3, u32 arg4); > +int meson_sm_call_read(void *buffer, unsigned int bsize, unsigned int cmd_index, > + u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4); > > #endif /* _MESON_SM_FW_H_ */ > ^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 2/2] firmware: meson-sm: Allow 0 as valid return value 2017-03-03 15:17 [PATCH 0/2] meson-sm: Fix issues when reading data from sm Carlo Caione 2017-03-03 15:17 ` [PATCH 1/2] firmware: meson-sm: Check for buffer output size Carlo Caione @ 2017-03-03 15:17 ` Carlo Caione 2017-03-22 15:55 ` [PATCH 0/2] meson-sm: Fix issues when reading data from sm Kevin Hilman 2 siblings, 0 replies; 6+ messages in thread From: Carlo Caione @ 2017-03-03 15:17 UTC (permalink / raw) To: linux-arm-kernel From: Carlo Caione <carlo@endlessm.com> Some special SMC calls (i.e. the function used to retrieve the serial number of the Amlogic SoCs) returns 0 in the register 0 also when the data was successfully read instead of using the register to hold the number of bytes returned in the bounce buffer as expected. With the current implementation of the driver this is seen as an error and meson_sm_call_read() returns an error even though the data was correctly read. To deal with this when we have no information about the amount of read data (that is 0 is returned by the SMC call) we return to the caller the requested amount of data and 0 as return value. Signed-off-by: Carlo Caione <carlo@endlessm.com> --- drivers/firmware/meson/meson_sm.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/drivers/firmware/meson/meson_sm.c b/drivers/firmware/meson/meson_sm.c index 5f30a5774e57..ff204421117b 100644 --- a/drivers/firmware/meson/meson_sm.c +++ b/drivers/firmware/meson/meson_sm.c @@ -136,11 +136,14 @@ EXPORT_SYMBOL(meson_sm_call); * @arg4: SMC32 Argument 4 * * Return: size of read data on success, a negative value on error + * When 0 is returned there is no guarantee about the amount of + * data read and bsize bytes are copied in buffer. */ int meson_sm_call_read(void *buffer, unsigned int bsize, unsigned int cmd_index, u32 arg0, u32 arg1, u32 arg2, u32 arg3, u32 arg4) { u32 size; + int ret; if (!fw.chip) return -ENOENT; @@ -154,13 +157,18 @@ int meson_sm_call_read(void *buffer, unsigned int bsize, unsigned int cmd_index, if (meson_sm_call(cmd_index, &size, arg0, arg1, arg2, arg3, arg4) < 0) return -EINVAL; - if (!size || size > bsize) + if (size > bsize) return -EINVAL; + ret = size; + + if (!size) + size = bsize; + if (buffer) memcpy(buffer, fw.sm_shmem_out_base, size); - return size; + return ret; } EXPORT_SYMBOL(meson_sm_call_read); -- 2.12.0 ^ permalink raw reply related [flat|nested] 6+ messages in thread
* [PATCH 0/2] meson-sm: Fix issues when reading data from sm 2017-03-03 15:17 [PATCH 0/2] meson-sm: Fix issues when reading data from sm Carlo Caione 2017-03-03 15:17 ` [PATCH 1/2] firmware: meson-sm: Check for buffer output size Carlo Caione 2017-03-03 15:17 ` [PATCH 2/2] firmware: meson-sm: Allow 0 as valid return value Carlo Caione @ 2017-03-22 15:55 ` Kevin Hilman 2017-03-23 10:31 ` Mark Rutland 2 siblings, 1 reply; 6+ messages in thread From: Kevin Hilman @ 2017-03-22 15:55 UTC (permalink / raw) To: linux-arm-kernel Carlo Caione <carlo@caione.org> writes: > From: Carlo Caione <carlo@endlessm.com> > > While working on a related driver I noticed a couple of problems in the > meson-sm driver. The problems are unrelated but since the fixes touch the same > codebase I decided to submit these in the same patchset. Applied to amlogic tree, branch: v4.12/drivers I'd prefer to have acks from Mark and Srini, but since this has been on the list for awhile, I'm guessing that isn't going to happen, so consider this a request. Kevin ^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH 0/2] meson-sm: Fix issues when reading data from sm 2017-03-22 15:55 ` [PATCH 0/2] meson-sm: Fix issues when reading data from sm Kevin Hilman @ 2017-03-23 10:31 ` Mark Rutland 0 siblings, 0 replies; 6+ messages in thread From: Mark Rutland @ 2017-03-23 10:31 UTC (permalink / raw) To: linux-arm-kernel On Wed, Mar 22, 2017 at 08:55:29AM -0700, Kevin Hilman wrote: > Carlo Caione <carlo@caione.org> writes: > > > From: Carlo Caione <carlo@endlessm.com> > > > > While working on a related driver I noticed a couple of problems in the > > meson-sm driver. The problems are unrelated but since the fixes touch the same > > codebase I decided to submit these in the same patchset. > > Applied to amlogic tree, branch: v4.12/drivers > > I'd prefer to have acks from Mark and Srini, but since this has been on > the list for awhile, I'm guessing that isn't going to happen, so > consider this a request. >From a quick scan, this looks fine to me. FWIW: Acked-by: Mark Rutland <mark.rutland@arm.com> Mark. ^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2017-03-23 10:31 UTC | newest] Thread overview: 6+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2017-03-03 15:17 [PATCH 0/2] meson-sm: Fix issues when reading data from sm Carlo Caione 2017-03-03 15:17 ` [PATCH 1/2] firmware: meson-sm: Check for buffer output size Carlo Caione 2017-03-22 16:28 ` Srinivas Kandagatla 2017-03-03 15:17 ` [PATCH 2/2] firmware: meson-sm: Allow 0 as valid return value Carlo Caione 2017-03-22 15:55 ` [PATCH 0/2] meson-sm: Fix issues when reading data from sm Kevin Hilman 2017-03-23 10:31 ` Mark Rutland
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).