Updates to MHI channel handling
Subsystems (2)
v5 cover Updates to MHI channel handling
MHI specification shows a state machine with support for STOP channel command and the validity of certain state transitions. MHI host currently does not provide any mechanism to stop a channel and restart it without resetting it. There are also times when the device moves on to a different execution environment while client drivers on the host are unaware of it and still attempt to reset the channels facing unnecessary timeouts.
This series addresses the above areas to provide support for stopping an MHI channel, resuming it back, improved documentation and improving upon channel state machine handling in general.
This set of patches was tested on arm64 architecture.
v5:
-Added reviewed-by tags from Hemant I missed earlier
-Added patch to prevent kernel warnings on clearing channel context twice
v4:
-Updated commit text/descriptions and addressed checkpatch checks -Added context validity check before starting/stopping channels from new API -Added patch to clear channel context configuration after reset/unprepare
v3:
-Updated documentation for channel transfer APIs to highlight differences -Create separate patch for "allowing channel to be disabled from stopped state"
v2:
-Renamed the newly introduced APIs to mhi_start_transfer() / mhi_stop_transfer() -Added improved documentation to avoid confusion with the new APIs -Removed the __ prefix from mhi_unprepare_channel() API for consistency.
Bhaumik Bhatt (9):
bus: mhi: core: Allow sending the STOP channel command
bus: mhi: core: Allow channel to be disabled from stopped state
bus: mhi: core: Improvements to the channel handling state machine
bus: mhi: core: Clear configuration from channel context during reset
bus: mhi: core: Add support to stop or start channel data transfers
bus: mhi: core: Check channel execution environment before issuing
reset
bus: mhi: core: Remove __ prefix for MHI channel unprepare function
bus: mhi: Improve documentation on channel transfer setup APIs
bus: mhi: core: Do not clear channel context more than once
drivers/bus/mhi/core/init.c | 23 +++- drivers/bus/mhi/core/internal.h | 12 ++ drivers/bus/mhi/core/main.c | 238 ++++++++++++++++++++++++++++------------ include/linux/mhi.h | 45 +++++++- 4 files changed, 243 insertions(+), 75 deletions(-)
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 1/9 bus: mhi: core: Allow sending the STOP channel command
Add support to allow sending the STOP channel command. If a client driver would like to STOP a channel and have the device retain the channel context instead of issuing a RESET to it and clearing the context, this would provide support for it after the ability to send this command is exposed to clients.
--- drivers/bus/mhi/core/main.c | 5 +++++ 1 file changed, 5 insertions(+)
drivers/bus/mhi/core/main.c +5 -0
@@ -1224,6 +1224,11 @@ int mhi_send_cmd(struct mhi_controller *mhi_cntrl, cmd_tre->dword[0 ] =MHI_TRE_CMD_RESET_DWORD0 ; cmd_tre->dword[1 ] =MHI_TRE_CMD_RESET_DWORD1 (chan);break ; +case MHI_CMD_STOP_CHAN : + cmd_tre->ptr =MHI_TRE_CMD_STOP_PTR ; + cmd_tre->dword[0 ] =MHI_TRE_CMD_STOP_DWORD0 ; + cmd_tre->dword[1 ] =MHI_TRE_CMD_STOP_DWORD1 (chan); +break ;case MHI_CMD_START_CHAN : cmd_tre->ptr =MHI_TRE_CMD_START_PTR ; cmd_tre->dword[0 ] =MHI_TRE_CMD_START_DWORD0 ;
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 1/9] bus: mhi: core: Allow sending the STOP channel command
Quoted 8 lines · Bhaumik Bhatt
Add support to allow sending the STOP channel command. If a client driver would like to STOP a channel and have the device retain the channel context instead of issuing a RESET to it and clearing the context, this would provide support for it after the ability to send this command is exposed to clients.
Signed-off-by Bhaumik Bhatt <[email protected]>Reviewed-by Hemant Kumar <[email protected]>
Thanks,
Mani
Quoted 24 lines
--- drivers/bus/mhi/core/main.c | 5 +++++ 1 file changed, 5 insertions(+)drivers/bus/mhi/core/main.c +5
-0@@ -1224,6 +1224,11 @@ int mhi_send_cmd(struct mhi_controller *mhi_cntrl, cmd_tre->dword[0 ] =MHI_TRE_CMD_RESET_DWORD0 ; cmd_tre->dword[1 ] =MHI_TRE_CMD_RESET_DWORD1 (chan);break ; +case MHI_CMD_STOP_CHAN : + cmd_tre->ptr =MHI_TRE_CMD_STOP_PTR ; + cmd_tre->dword[0 ] =MHI_TRE_CMD_STOP_DWORD0 ; + cmd_tre->dword[1 ] =MHI_TRE_CMD_STOP_DWORD1 (chan); +break ;case MHI_CMD_START_CHAN : cmd_tre->ptr =MHI_TRE_CMD_START_PTR ; cmd_tre->dword[0 ] =MHI_TRE_CMD_START_DWORD0 ;Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 2/9 bus: mhi: core: Allow channel to be disabled from stopped state
In reply to Bhaumik BhattIf a channel was explicitly stopped but not reset, allow it to move to a disabled state so that the channel context can be cleaned up after a driver remove is issued. Since the channel remained in stopped state, its context on the device is not cleared. Allow this move if a client driver module is unloaded or a device crash occurs.
--- drivers/bus/mhi/core/init.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/bus/mhi/core/init.c +2 -1
@@ -1293,7 +1293,8 @@ static int mhi_driver_remove(struct device *dev)mutex_lock (&mhi_chan->mutex); -if (ch_state[dir] ==MHI_CH_STATE_ENABLED && +if ((ch_state[dir] ==MHI_CH_STATE_ENABLED || + ch_state[dir] ==MHI_CH_STATE_STOP ) && !mhi_chan->offload_ch)mhi_deinit_chan_ctxt (mhi_cntrl, mhi_chan);
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 2/9] bus: mhi: core: Allow channel to be disabled from stopped state
Quoted 7 lines · Bhaumik Bhatt
If a channel was explicitly stopped but not reset, allow it to move to a disabled state so that the channel context can be cleaned up after a driver remove is issued. Since the channel remained in stopped state, its context on the device is not cleared. Allow this move if a client driver module is unloaded or a device crash occurs.
The commit message needs to be improved. The stopped channel will by default move to stopped state during mhi_driver_remove(). So this patch is not at all doing anything other than clearing the context for stopped channels.
Please fix it!
Quoted 2 lines
Signed-off-by Bhaumik Bhatt <[email protected]>Reviewed-by Hemant Kumar <[email protected]>
Thanks,
Mani
Quoted 22 lines
--- drivers/bus/mhi/core/init.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)drivers/bus/mhi/core/init.c +2
-1@@ -1293,7 +1293,8 @@ static int mhi_driver_remove(struct device *dev)mutex_lock (&mhi_chan->mutex); -if (ch_state[dir] ==MHI_CH_STATE_ENABLED && +if ((ch_state[dir] ==MHI_CH_STATE_ENABLED || + ch_state[dir] ==MHI_CH_STATE_STOP ) && !mhi_chan->offload_ch)mhi_deinit_chan_ctxt (mhi_cntrl, mhi_chan);Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 3/9 bus: mhi: core: Improvements to the channel handling state machine
In reply to Bhaumik BhattImprove the channel handling state machine such that all commands go through a common function and validation process to ensure that the state machine is not violated in any way and adheres to the MHI specification.
--- drivers/bus/mhi/core/init.c | 6 ++ drivers/bus/mhi/core/internal.h | 12 +++ drivers/bus/mhi/core/main.c | 166 +++++++++++++++++++++++----------------- 3 files changed, 114 insertions(+), 70 deletions(-)
drivers/bus/mhi/core/init.c +6 -0
@@ -54,6 +54,12 @@ const char * const mhi_state_str[MHI_STATE_MAX] = { [MHI_STATE_SYS_ERR] = "SYS_ERR", }; +const char *const mhi_ch_state_type_str[MHI_CH_STATE_TYPE_MAX ] = { + [MHI_CH_STATE_TYPE_RESET ] ="RESET" , + [MHI_CH_STATE_TYPE_STOP ] ="STOP" , + [MHI_CH_STATE_TYPE_START ] ="START" , +}; +static const char *const mhi_pm_state_str[] = { [MHI_PM_STATE_DISABLE ] ="DISABLE" , [MHI_PM_STATE_POR ] ="POR" ,
drivers/bus/mhi/core/internal.h +12 -0
@@ -369,6 +369,18 @@ enum mhi_ch_state { MHI_CH_STATE_ERROR = 0x5, }; +enum mhi_ch_state_type { +MHI_CH_STATE_TYPE_RESET , +MHI_CH_STATE_TYPE_STOP , +MHI_CH_STATE_TYPE_START , +MHI_CH_STATE_TYPE_MAX , +}; + +extern const char *const mhi_ch_state_type_str[MHI_CH_STATE_TYPE_MAX ]; +#define TO_CH_STATE_TYPE_STR (state) (((state) >= MHI_CH_STATE_TYPE_MAX) ? \ + "INVALID_STATE" : \ + mhi_ch_state_type_str[(state)]) +#define MHI_INVALID_BRSTMODE (mode) (mode != MHI_DB_BRST_DISABLE && \ mode != MHI_DB_BRST_ENABLE)
drivers/bus/mhi/core/main.c +96 -70
217 lines not on this page · raw message
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 3/9] bus: mhi: core: Improvements to the channel handling state machine
Quoted 64 lines · Bhaumik Bhatt
Improve the channel handling state machine such that all commands go through a common function and validation process to ensure that the state machine is not violated in any way and adheres to the MHI specification.
Signed-off-by Bhaumik Bhatt <[email protected]>Reviewed-by Hemant Kumar <[email protected]>--- drivers/bus/mhi/core/init.c | 6 ++ drivers/bus/mhi/core/internal.h | 12 +++ drivers/bus/mhi/core/main.c | 166 +++++++++++++++++++++++----------------- 3 files changed, 114 insertions(+), 70 deletions(-)drivers/bus/mhi/core/init.c +6
-0@@ -54,6 +54,12 @@ const char * const mhi_state_str[MHI_STATE_MAX] = { [MHI_STATE_SYS_ERR] = "SYS_ERR", }; +const char *const mhi_ch_state_type_str[MHI_CH_STATE_TYPE_MAX ] = { + [MHI_CH_STATE_TYPE_RESET ] ="RESET" , + [MHI_CH_STATE_TYPE_STOP ] ="STOP" , + [MHI_CH_STATE_TYPE_START ] ="START" , +}; +static const char *const mhi_pm_state_str[] = { [MHI_PM_STATE_DISABLE ] ="DISABLE" , [MHI_PM_STATE_POR ] ="POR" ,drivers/bus/mhi/core/internal.h +12
-0@@ -369,6 +369,18 @@ enum mhi_ch_state { MHI_CH_STATE_ERROR = 0x5, }; +enum mhi_ch_state_type { +MHI_CH_STATE_TYPE_RESET , +MHI_CH_STATE_TYPE_STOP , +MHI_CH_STATE_TYPE_START , +MHI_CH_STATE_TYPE_MAX , +}; + +extern const char *const mhi_ch_state_type_str[MHI_CH_STATE_TYPE_MAX ]; +#define TO_CH_STATE_TYPE_STR (state) (((state) >= MHI_CH_STATE_TYPE_MAX) ? \ + "INVALID_STATE" : \ + mhi_ch_state_type_str[(state)]) +#define MHI_INVALID_BRSTMODE (mode) (mode != MHI_DB_BRST_DISABLE && \ mode != MHI_DB_BRST_ENABLE)drivers/bus/mhi/core/main.c +1
-2@@ -1250,56 +1250,118 @@ int mhi_send_cmd(struct mhi_controller *mhi_cntrl,return 0 ; } -static void __mhi_unprepare_channel (struct mhi_controller *mhi_cntrl, -struct mhi_chan *mhi_chan) +static int mhi_update_channel_state(struct mhi_controller *mhi_cntrl,
drivers/bus/mhi/core/main.c int mhi_send_cmd(struct mhi_controller *mhi_cntrl, · line 1262
Quoted 10 lines
drivers/bus/mhi/core/main.c +7
-1+struct mhi_chan *mhi_chan, + enum mhi_ch_state_type to_state) { -int ret; struct device *dev = &mhi_cntrl->mhi_dev->dev; +enum mhi_cmd_type cmd =MHI_CMD_NOP ; +int ret = -EIO ; + +dev_dbg (dev,"Updating channel %s(%d) state to: %s\n" , mhi_chan->name, + mhi_chan->chan,TO_CH_STATE_TYPE_STR (to_state));
Instead of explicitly using the channel name and its id, you should be able to use mhi_chan->mhi_dev->dev so that dev_dbg() can provide that information for you. For instance,
mhi0_IPCR: Updating channel state to RESET
Do it for all instances below.
Quoted 35 lines
drivers/bus/mhi/core/main.c +24
-9+ +switch (to_state) { +case MHI_CH_STATE_TYPE_RESET : +write_lock_irq (&mhi_chan->lock); +if (mhi_chan->ch_state !=MHI_CH_STATE_STOP && + mhi_chan->ch_state !=MHI_CH_STATE_ENABLED && + mhi_chan->ch_state !=MHI_CH_STATE_SUSPENDED ) { +write_unlock_irq (&mhi_chan->lock); + goto exit_invalid_state; + } + mhi_chan->ch_state =MHI_CH_STATE_DISABLED ; +write_unlock_irq (&mhi_chan->lock); -dev_dbg (dev,"Entered: unprepare channel:%d\n" , mhi_chan->chan); + cmd =MHI_CMD_RESET_CHAN ; +break ; +case MHI_CH_STATE_TYPE_STOP : +if (mhi_chan->ch_state !=MHI_CH_STATE_ENABLED ) + goto exit_invalid_state; -/* no more processing events for this channel */ -mutex_lock (&mhi_chan->mutex); -write_lock_irq (&mhi_chan->lock); -if (mhi_chan->ch_state !=MHI_CH_STATE_ENABLED && - mhi_chan->ch_state !=MHI_CH_STATE_SUSPENDED ) { -write_unlock_irq (&mhi_chan->lock); -mutex_unlock (&mhi_chan->mutex); -return ; + cmd =MHI_CMD_STOP_CHAN ; +break ; +case MHI_CH_STATE_TYPE_START : +if (mhi_chan->ch_state !=MHI_CH_STATE_STOP && + mhi_chan->ch_state !=MHI_CH_STATE_DISABLED ) + goto exit_invalid_state; +
drivers/bus/mhi/core/main.c int mhi_send_cmd(struct mhi_controller *mhi_cntrl, · line 1296
Quoted 10 lines
drivers/bus/mhi/core/main.c +6
-2+ cmd =MHI_CMD_START_CHAN ; +break ; +default : + goto exit_invalid_state; } - mhi_chan->ch_state =MHI_CH_STATE_DISABLED ; -write_unlock_irq (&mhi_chan->lock); +/* bring host and device out of suspended states */ + ret =mhi_device_get_sync (mhi_cntrl->mhi_dev);
You are sneaking in a change here. Please add it to the commit description.
Quoted 47 lines
drivers/bus/mhi/core/main.c +31
-10+if (ret) +return ret; + mhi_cntrl->runtime_get (mhi_cntrl);reinit_completion (&mhi_chan->completion); -read_lock_bh (&mhi_cntrl->pm_lock); -if (MHI_PM_IN_ERROR_STATE (mhi_cntrl->pm_state)) { -read_unlock_bh (&mhi_cntrl->pm_lock); - goto error_invalid_state; + ret =mhi_send_cmd (mhi_cntrl, mhi_chan, cmd); +if (ret) { +dev_err (dev,"Failed to send %s(%d) %s command\n" , + mhi_chan->name, mhi_chan->chan, +TO_CH_STATE_TYPE_STR (to_state)); + goto exit_command_failure; } - mhi_cntrl->wake_toggle (mhi_cntrl); -read_unlock_bh (&mhi_cntrl->pm_lock); + ret =wait_for_completion_timeout (&mhi_chan->completion, +msecs_to_jiffies (mhi_cntrl->timeout_ms)); +if (!ret || mhi_chan->ccs !=MHI_EV_CC_SUCCESS ) { +dev_err (dev,"Failed to receive %s(%d) %s command completion\n" , + mhi_chan->name, mhi_chan->chan, +TO_CH_STATE_TYPE_STR (to_state)); + ret = -EIO ; + goto exit_command_failure; + } - mhi_cntrl->runtime_get (mhi_cntrl); + ret =0 ; + +if (to_state !=MHI_CH_STATE_TYPE_RESET ) { +write_lock_irq (&mhi_chan->lock); + mhi_chan->ch_state = (to_state ==MHI_CH_STATE_TYPE_START ) ? +MHI_CH_STATE_ENABLED :MHI_CH_STATE_STOP ; +write_unlock_irq (&mhi_chan->lock); + } + +dev_dbg (dev,"Channel %s(%d) state change to %s successful\n" , + mhi_chan->name, mhi_chan->chan,TO_CH_STATE_TYPE_STR (to_state)); + +exit_command_failure: mhi_cntrl->runtime_put (mhi_cntrl); - ret =mhi_send_cmd (mhi_cntrl, mhi_chan,MHI_CMD_RESET_CHAN ); -if (ret) - goto error_invalid_state;
drivers/bus/mhi/core/main.c int mhi_send_cmd(struct mhi_controller *mhi_cntrl, · line 1337
Quoted 10 lines
drivers/bus/mhi/core/main.c +3
-6+mhi_device_put (mhi_cntrl->mhi_dev); -/* even if it fails we will still reset */ - ret =wait_for_completion_timeout (&mhi_chan->completion, -msecs_to_jiffies (mhi_cntrl->timeout_ms)); -if (!ret || mhi_chan->ccs !=MHI_EV_CC_SUCCESS ) -dev_err (dev, -"Failed to receive cmd completion, still resetting\n" ); +return ret; +
This is not a recommended way to handle error paths. There should be a single return for all error paths. Please fix it.
Quoted 4 lines
drivers/bus/mhi/core/main.c +4
-0+exit_invalid_state: +dev_err (dev,"Channel %s(%d) update to %s not allowed\n" , + mhi_chan->name, mhi_chan->chan,TO_CH_STATE_TYPE_STR (to_state)); +
drivers/bus/mhi/core/main.c int mhi_send_cmd(struct mhi_controller *mhi_cntrl, · line 1351
Quoted 10 lines
drivers/bus/mhi/core/main.c +10
-0+return -EINVAL ; +} + +static void __mhi_unprepare_channel (struct mhi_controller *mhi_cntrl, +struct mhi_chan *mhi_chan) +{ +int ret; +struct device *dev = &mhi_cntrl->mhi_dev->dev; + +/* no more processing events for this channel */
Move this comment below.
Thanks,
Mani
PATCH v5 4/9 bus: mhi: core: Clear configuration from channel context during reset
In reply to Bhaumik BhattWhen clearing up the channel context after client drivers are done using channels, the configuration is currently not being reset entirely. Ensure this is done to appropriately handle issues where clients unaware of the context state end up calling functions which expect a context.
--- drivers/bus/mhi/core/init.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)
drivers/bus/mhi/core/init.c +13 -0
@@ -558,6 +558,7 @@ void mhi_deinit_chan_ctxt(struct mhi_controller *mhi_cntrl,struct mhi_ring *buf_ring;struct mhi_ring *tre_ring;struct mhi_chan_ctxt *chan_ctxt; +u32 tmp; buf_ring = &mhi_chan->buf_ring; tre_ring = &mhi_chan->tre_ring;
@@ -568,7 +569,19 @@ void mhi_deinit_chan_ctxt(struct mhi_controller *mhi_cntrl,vfree (buf_ring->base); buf_ring->base = tre_ring->base =NULL ; + tre_ring->ctxt_wp =NULL ; chan_ctxt->rbase =0 ; + chan_ctxt->rlen =0 ; + chan_ctxt->rp =0 ; + chan_ctxt->wp =0 ; + + tmp = chan_ctxt->chcfg; + tmp &= ~CHAN_CTX_CHSTATE_MASK ; + tmp |= (MHI_CH_STATE_DISABLED <<CHAN_CTX_CHSTATE_SHIFT ); + chan_ctxt->chcfg = tmp; + +/* Update to all cores */ +smp_wmb (); } int mhi_init_chan_ctxt(struct mhi_controller *mhi_cntrl,
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 4/9] bus: mhi: core: Clear configuration from channel context during reset
Quoted 8 lines · Bhaumik Bhatt
When clearing up the channel context after client drivers are done using channels, the configuration is currently not being reset entirely. Ensure this is done to appropriately handle issues where clients unaware of the context state end up calling functions which expect a context.
Signed-off-by Bhaumik Bhatt <[email protected]>Reviewed-by Hemant Kumar <[email protected]>
Thanks,
Mani
Quoted 40 lines
--- drivers/bus/mhi/core/init.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)drivers/bus/mhi/core/init.c +13
-0@@ -558,6 +558,7 @@ void mhi_deinit_chan_ctxt(struct mhi_controller *mhi_cntrl,struct mhi_ring *buf_ring;struct mhi_ring *tre_ring;struct mhi_chan_ctxt *chan_ctxt; +u32 tmp; buf_ring = &mhi_chan->buf_ring; tre_ring = &mhi_chan->tre_ring;@@ -568,7 +569,19 @@ void mhi_deinit_chan_ctxt(struct mhi_controller *mhi_cntrl,vfree (buf_ring->base); buf_ring->base = tre_ring->base =NULL ; + tre_ring->ctxt_wp =NULL ; chan_ctxt->rbase =0 ; + chan_ctxt->rlen =0 ; + chan_ctxt->rp =0 ; + chan_ctxt->wp =0 ; + + tmp = chan_ctxt->chcfg; + tmp &= ~CHAN_CTX_CHSTATE_MASK ; + tmp |= (MHI_CH_STATE_DISABLED <<CHAN_CTX_CHSTATE_SHIFT ); + chan_ctxt->chcfg = tmp; + +/* Update to all cores */ +smp_wmb (); } int mhi_init_chan_ctxt(struct mhi_controller *mhi_cntrl,Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 5/9 bus: mhi: core: Add support to stop or start channel data transfers
In reply to Bhaumik BhattSome MHI client drivers may want to request a pause or halt of data transfer activity on their channels. Support for this does not exist and must be introduced, wherein the channel context is not reset or cleared but only the STOP channel command is issued. This would need to be paired with an API that allows resuming the data transfer activity on channels by use of the START channel command. This API assumes that the context information is already setup. Enable this using two new APIs, mhi_start_transfer() and mhi_stop_transfer().
--- drivers/bus/mhi/core/main.c | 50 +++++++++++++++++++++++++++++++++++++++++++++ include/linux/mhi.h | 19 +++++++++++++++++ 2 files changed, 69 insertions(+)
drivers/bus/mhi/core/main.c +50 -0
@@ -1583,6 +1583,56 @@ void mhi_unprepare_from_transfer(struct mhi_device *mhi_dev) }EXPORT_SYMBOL_GPL (mhi_unprepare_from_transfer); +static int mhi_update_transfer_state (struct mhi_device *mhi_dev, +enum mhi_ch_state_type to_state) +{ +struct mhi_controller *mhi_cntrl = mhi_dev->mhi_cntrl; +struct device *dev = &mhi_cntrl->mhi_dev->dev; +struct mhi_chan *mhi_chan; +struct mhi_chan_ctxt *chan_ctxt; +int dir, ret; + +for (dir =0 ; dir <2 ; dir++) { + mhi_chan = dir ? mhi_dev->ul_chan : mhi_dev->dl_chan; + +if (!mhi_chan) +continue ; + +/* +* Bail out if one of the channels fails as client will reset +* both upon failure +*/ +mutex_lock (&mhi_chan->mutex); + chan_ctxt = &mhi_cntrl->mhi_ctxt->chan_ctxt[mhi_chan->chan]; +if (!(chan_ctxt->chcfg &CHAN_CTX_CHSTATE_MASK )) { +mutex_unlock (&mhi_chan->mutex); +dev_err (dev,"Channel %s(%u) context not initialized\n" , + mhi_chan->name, mhi_chan->chan); +return -EINVAL ; + } + ret =mhi_update_channel_state (mhi_cntrl, mhi_chan, to_state); +if (ret) { +mutex_unlock (&mhi_chan->mutex); +return ret; + } +mutex_unlock (&mhi_chan->mutex); + } + +return 0 ; +} + +int mhi_stop_transfer (struct mhi_device *mhi_dev) +{ +return mhi_update_transfer_state (mhi_dev,MHI_CH_STATE_TYPE_STOP ); +} +EXPORT_SYMBOL_GPL (mhi_stop_transfer); + +int mhi_start_transfer (struct mhi_device *mhi_dev) +{ +return mhi_update_transfer_state (mhi_dev,MHI_CH_STATE_TYPE_START ); +} +EXPORT_SYMBOL_GPL (mhi_start_transfer); +int mhi_poll (struct mhi_device *mhi_dev,u32 budget) {struct mhi_controller *mhi_cntrl = mhi_dev->mhi_cntrl;
include/linux/mhi.h +19 -0
@@ -715,6 +715,25 @@ int mhi_prepare_for_transfer(struct mhi_device *mhi_dev);void mhi_unprepare_from_transfer (struct mhi_device *mhi_dev);/** +* mhi_stop_transfer - Pauses ongoing channel activity by issuing the STOP +* channel command to both UL and DL channels. This command +* does not reset the channel context and the client drivers +* can issue mhi_start_transfer to resume activity. +* @mhi_dev: Device associated with the channels +*/ +int mhi_stop_transfer (struct mhi_device *mhi_dev); + +/** +* mhi_start_transfer - Resumes channel activity by issuing the START channel +* command to both UL and DL channels. This command assumes +* the channel context is already setup and the client +* drivers can issue mhi_stop_transfer to pause activity if +* required. +* @mhi_dev: Device associated with the channels +*/ +int mhi_start_transfer (struct mhi_device *mhi_dev); + +/** * mhi_poll - Poll for any available data in DL direction * @mhi_dev: Device associated with the channels * @budget: # of events to process
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 5/9] bus: mhi: core: Add support to stop or start channel data transfers
Quoted 10 lines · Bhaumik Bhatt
Some MHI client drivers may want to request a pause or halt of data transfer activity on their channels. Support for this does not exist and must be introduced, wherein the channel context is not reset or cleared but only the STOP channel command is issued. This would need to be paired with an API that allows resuming the data transfer activity on channels by use of the START channel command. This API assumes that the context information is already setup. Enable this using two new APIs,
mhi_start_transfer()andmhi_stop_transfer().
Do you have any user for these APIs? If not then please post it when one is available.
Thanks,
Mani
Re: [PATCH v5 5/9] bus: mhi: core: Add support to stop or start channel data transfers
Quoted 12 lines · Manivannan Sadhasivam
Quoted 10 lines · Bhaumik Bhatt
Some MHI client drivers may want to request a pause or halt of data transfer activity on their channels. Support for this does not exist and must be introduced, wherein the channel context is not reset or cleared but only the STOP channel command is issued. This would need to be paired with an API that allows resuming the data transfer activity on channels by use of the START channel command. This API assumes that the context information is already setup. Enable this using two new APIs,
mhi_start_transfer()andmhi_stop_transfer().
Quoted 5 lines
Do you have any user for these APIs? If not then please post it when one is available.
Thanks,
Mani
I will be dropping this patch in upcoming v6 series.
Thanks,
Bhaumik
---
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 6/9 bus: mhi: core: Check channel execution environment before issuing reset
In reply to Bhaumik BhattA client can attempt to unprepare certain channels for transfer even after the execution environment they are supposed to run in has changed. In the event that happens, the device need not be notified of the reset and the host can proceed with clean up for the channel context and memory allocated for it on the host as the device will no longer be able to respond to such a request.
--- drivers/bus/mhi/core/main.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)
drivers/bus/mhi/core/main.c +13 -0
@@ -1351,11 +1351,24 @@ static void __mhi_unprepare_channel(struct mhi_controller *mhi_cntrl,/* no more processing events for this channel */ mutex_lock (&mhi_chan->mutex); +if (!(BIT (mhi_cntrl->ee) & mhi_chan->ee_mask)) { +dev_err (dev, +"Current EE: %s Required EE Mask: 0x%x for chan: %s\n" , +TO_MHI_EXEC_STR (mhi_cntrl->ee), mhi_chan->ee_mask, + mhi_chan->name); + goto exit_unprepare_channel; + } + ret =mhi_update_channel_state (mhi_cntrl, mhi_chan,MHI_CH_STATE_TYPE_RESET );if (ret)dev_err (dev,"Failed to reset channel, still resetting\n" ); +exit_unprepare_channel: +write_lock_irq (&mhi_chan->lock); + mhi_chan->ch_state =MHI_CH_STATE_DISABLED ; +write_unlock_irq (&mhi_chan->lock); +if (!mhi_chan->offload_ch) {mhi_reset_chan (mhi_cntrl, mhi_chan);mhi_deinit_chan_ctxt (mhi_cntrl, mhi_chan);
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 6/9] bus: mhi: core: Check channel execution environment before issuing reset
Quoted 2 lines · Bhaumik Bhatt
A client can attempt to unprepare certain channels for transfer even after the execution environment they are supposed to run in has changed.
Just out of curiosity, under what circumstances it can happen? We don't have any in kernel user for the mhi_unprepare_from_transfer() API :/
Quoted 16 lines
In the event that happens, the device need not be notified of the reset and the host can proceed with clean up for the channel context and memory allocated for it on the host as the device will no longer be able to respond to such a request.
Signed-off-by Bhaumik Bhatt <[email protected]>Reviewed-by Hemant Kumar <[email protected]>--- drivers/bus/mhi/core/main.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)drivers/bus/mhi/core/main.c +0
-0
drivers/bus/mhi/core/main.c static void __mhi_unprepare_channel(struct mhi_controller *mhi_cntrl, · line 1358
Quoted 8 lines
drivers/bus/mhi/core/main.c +5
-0@@ -1351,11 +1351,24 @@ static void __mhi_unprepare_channel(struct mhi_controller *mhi_cntrl,/* no more processing events for this channel */ mutex_lock (&mhi_chan->mutex); +if (!(BIT (mhi_cntrl->ee) & mhi_chan->ee_mask)) { +dev_err (dev, +"Current EE: %s Required EE Mask: 0x%x for chan: %s\n" , +TO_MHI_EXEC_STR (mhi_cntrl->ee), mhi_chan->ee_mask, + mhi_chan->name);
Again, use channel's struct dev here.
Thanks,
Mani
Re: [PATCH v5 6/9] bus: mhi: core: Check channel execution environment before issuing reset
Hi Mani,
Quoted 7 lines · Manivannan Sadhasivam
Quoted 2 lines · Bhaumik Bhatt
A client can attempt to unprepare certain channels for transfer even after the execution environment they are supposed to run in has changed.
Just out of curiosity, under what circumstances it can happen? We don't have any in kernel user for the
mhi_unprepare_from_transfer()API :/
We do and it's the qrtr-mhi driver where we can potentially see this happen. Mainly, since they want to undo whatever they have done in probe(), this can come from the remove() after MHI power down is initiated.
I have updated this patch in upcoming v6 series with a debug message as we only want to proceed with clean-up directly.
Quoted 20 lines
Quoted 20 lines
In the event that happens, the device need not be notified of the reset and the host can proceed with clean up for the channel context and memory allocated for it on the host as the device will no longer be able to respond to such a request.
Signed-off-by Bhaumik Bhatt <[email protected]>Reviewed-by Hemant Kumar <[email protected]>--- drivers/bus/mhi/core/main.c | 13 +++++++++++++ 1 file changed, 13 insertions(+)drivers/bus/mhi/core/main.c +0
-0@@ -1351,11 +1351,24 @@ static void __mhi_unprepare_channel(struct mhi_controller *mhi_cntrl,/* no more processing events for this channel */ mutex_lock (&mhi_chan->mutex);
Quoted 8 lines
drivers/bus/mhi/core/main.c static void __mhi_unprepare_channel(struct mhi_controller *mhi_cntrl, · line 1358
Quoted 5 lines
drivers/bus/mhi/core/main.c +5
-0+if (!(BIT (mhi_cntrl->ee) & mhi_chan->ee_mask)) { +dev_err (dev, +"Current EE: %s Required EE Mask: 0x%x for chan: %s\n" , +TO_MHI_EXEC_STR (mhi_cntrl->ee), mhi_chan->ee_mask, + mhi_chan->name);Again, use channel's struct dev here.
Done.
Quoted 2 lines
Thanks,
Mani
Thanks,
Bhaumik
---
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 7/9 bus: mhi: core: Remove __ prefix for MHI channel unprepare function
In reply to Bhaumik BhattThe __mhi_unprepare_channel() API does not require the __ prefix. Get rid of it and make the internal function consistent with the other function names.
--- drivers/bus/mhi/core/main.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/bus/mhi/core/main.c +5 -5
@@ -1342,8 +1342,8 @@ static int mhi_update_channel_state(struct mhi_controller *mhi_cntrl,return -EINVAL ; } -static void __mhi_unprepare_channel (struct mhi_controller *mhi_cntrl, -struct mhi_chan *mhi_chan) +static void mhi_unprepare_channel (struct mhi_controller *mhi_cntrl, +struct mhi_chan *mhi_chan) {int ret;struct device *dev = &mhi_cntrl->mhi_dev->dev;
@@ -1456,7 +1456,7 @@ int mhi_prepare_channel(struct mhi_controller *mhi_cntrl, error_pre_alloc:mutex_unlock (&mhi_chan->mutex); -__mhi_unprepare_channel (mhi_cntrl, mhi_chan); +mhi_unprepare_channel (mhi_cntrl, mhi_chan);return ret; }
@@ -1573,7 +1573,7 @@ int mhi_prepare_for_transfer(struct mhi_device *mhi_dev)if (!mhi_chan)continue ; -__mhi_unprepare_channel (mhi_cntrl, mhi_chan); +mhi_unprepare_channel (mhi_cntrl, mhi_chan); }return ret;
@@ -1591,7 +1591,7 @@ void mhi_unprepare_from_transfer(struct mhi_device *mhi_dev)if (!mhi_chan)continue ; -__mhi_unprepare_channel (mhi_cntrl, mhi_chan); +mhi_unprepare_channel (mhi_cntrl, mhi_chan); } }EXPORT_SYMBOL_GPL (mhi_unprepare_from_transfer);
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 8/9 bus: mhi: Improve documentation on channel transfer setup APIs
In reply to Bhaumik BhattThe mhi_prepare_for_transfer() and mhi_unprepare_from_transfer() APIs could use better explanation, especially with the addition of two new APIs to start and stop the transfers on channels. Add better set of information for those APIs.
--- include/linux/mhi.h | 26 ++++++++++++++++++++++++-- 1 file changed, 24 insertions(+), 2 deletions(-)
include/linux/mhi.h +24 -2
@@ -703,13 +703,35 @@ int mhi_device_get_sync(struct mhi_device *mhi_dev);void mhi_device_put (struct mhi_device *mhi_dev);/** -* mhi_prepare_for_transfer - Setup channel for data transfer +* mhi_prepare_for_transfer - Setup UL and DL channels for data transfer. +* Allocate and initialize the channel context and +* also issue the START channel command to both +* channels. Channels can be started only if both +* host and device execution environments match and +* channels are in a DISABLED state. Calling the +* mhi_start_transfer() function is not required +* afterwards as channels are already started. This +* function also initializes the channel context +* whereas mhi_start_transfer() can only be used to +* issue the start channel command once the context +* is setup. * @mhi_dev: Device associated with the channels */ int mhi_prepare_for_transfer (struct mhi_device *mhi_dev);/** -* mhi_unprepare_from_transfer - Unprepare the channels +* mhi_unprepare_from_transfer - Reset UL and DL channels for data transfer. +* Issue the RESET channel command and let the +* device clean-up the context so no incoming +* transfers are seen on the host. Free memory +* associated with the context on host. If device +* is unresponsive, only perform a host side +* clean-up. Channels can be reset only if both +* host and device execution environments match +* and channels are in an ENABLED, STOPPED or +* SUSPENDED state. Calling mhi_stop_transfer() is +* not required before calling this function as it +* will only stop transfers, not reset channels. * @mhi_dev: Device associated with the channels */ void mhi_unprepare_from_transfer (struct mhi_device *mhi_dev);
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
PATCH v5 9/9 bus: mhi: core: Do not clear channel context more than once
In reply to Bhaumik BhattWhen clearing the channel context, calling mhi_free_coherent() more than once can result in kernel warnings such as "trying to free invalid coherent area". Prevent extra work by adding a check to skip calling mhi_deinit_chan_ctxt() if the client driver has already disabled the channels.
--- drivers/bus/mhi/core/init.c | 1 + 1 file changed, 1 insertion(+)
drivers/bus/mhi/core/init.c +1 -0
@@ -1314,6 +1314,7 @@ static int mhi_driver_remove(struct device *dev)if ((ch_state[dir] ==MHI_CH_STATE_ENABLED || ch_state[dir] ==MHI_CH_STATE_STOP ) && + mhi_chan->ch_state !=MHI_CH_STATE_DISABLED && !mhi_chan->offload_ch)mhi_deinit_chan_ctxt (mhi_cntrl, mhi_chan);
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 9/9] bus: mhi: core: Do not clear channel context more than once
Quoted 17 lines · Bhaumik Bhatt
When clearing the channel context, calling
mhi_free_coherent()more than once can result in kernel warnings such as "trying to free invalid coherent area". Prevent extra work by adding a check to skip callingmhi_deinit_chan_ctxt()if the client driver has already disabled the channels.Signed-off-by Bhaumik Bhatt <[email protected]>--- drivers/bus/mhi/core/init.c | 1 + 1 file changed, 1 insertion(+)drivers/bus/mhi/core/init.c +0
-0
drivers/bus/mhi/core/init.c static int mhi_driver_remove(struct device *dev) · line 1320
Quoted 7 lines
drivers/bus/mhi/core/init.c +1
-0@@ -1314,6 +1314,7 @@ static int mhi_driver_remove(struct device *dev)if ((ch_state[dir] ==MHI_CH_STATE_ENABLED || ch_state[dir] ==MHI_CH_STATE_STOP ) && + mhi_chan->ch_state !=MHI_CH_STATE_DISABLED && !mhi_chan->offload_ch)mhi_deinit_chan_ctxt (mhi_cntrl, mhi_chan);
Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project
Re: [PATCH v5 9/9] bus: mhi: core: Do not clear channel context more than once
In reply to Bhaumik BhattQuoted 6 lines · Bhaumik Bhatt
When clearing the channel context, calling
mhi_free_coherent()more than once can result in kernel warnings such as "trying to free invalid coherent area". Prevent extra work by adding a check to skip callingmhi_deinit_chan_ctxt()if the client driver has already disabled the channels.
Again, please explain where and when exactly you spotted this issue.
Thanks,
Mani
Quoted 21 lines
Signed-off-by Bhaumik Bhatt <[email protected]>--- drivers/bus/mhi/core/init.c | 1 + 1 file changed, 1 insertion(+)drivers/bus/mhi/core/init.c +1
-0@@ -1314,6 +1314,7 @@ static int mhi_driver_remove(struct device *dev)if ((ch_state[dir] ==MHI_CH_STATE_ENABLED || ch_state[dir] ==MHI_CH_STATE_STOP ) && + mhi_chan->ch_state !=MHI_CH_STATE_DISABLED && !mhi_chan->offload_ch)mhi_deinit_chan_ctxt (mhi_cntrl, mhi_chan);Signature · 2 lines
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project