Re: [PATCH 10/11] drivers: bus: mhi: host: Add support for MHI MAX TRB capability
Manivannan Sadhasivam <[email protected]>
| Newsgroups | dev.linux.lists.mhi,org.kernel.vger.linux-arm-msm,org.kernel.vger.linux-doc,org.kernel.vger.linux-kernel |
|---|---|
| Message-ID | <gok5wlsti3y7eebdo4rjbvoclo3d2o2e3bseo3lwrooo5sf64a@42balu3tsz5e> |
On Thu, Dec 11, 2025 at 01:37:42PM +0530, Sivareddy Surasani wrote: > From: Vivek Pernamitta <[email protected]> > > Read the MHI capability for MAX TRB length if device is supporting. Expand TRB. > Use this information to send MHI data with a higher TRB length, as > supported by the device. > Add some context on what the device is going to do with it. > Signed-off-by: Vivek Pernamitta <[email protected]> > Signed-off-by: Sivareddy Surasani <[email protected]> > --- > drivers/bus/mhi/common.h | 9 ++++++++- > drivers/bus/mhi/host/init.c | 21 +++++++++++++++++++++ > drivers/bus/mhi/host/main.c | 17 ++++++++++++++--- > include/linux/mhi.h | 2 ++ > 4 files changed, 45 insertions(+), 4 deletions(-) > > diff --git a/drivers/bus/mhi/common.h b/drivers/bus/mhi/common.h > index 3b3ecbc6169f..4962035f4693 100644 > --- a/drivers/bus/mhi/common.h > +++ b/drivers/bus/mhi/common.h > @@ -158,6 +158,8 @@ > #define MHI_TRE_GET_EV_PTR(tre) le64_to_cpu((tre)->ptr) > #define MHI_TRE_GET_EV_CODE(tre) FIELD_GET(GENMASK(31, 24), (MHI_TRE_GET_DWORD(tre, 0))) > #define MHI_TRE_GET_EV_LEN(tre) FIELD_GET(GENMASK(15, 0), (MHI_TRE_GET_DWORD(tre, 0))) > +#define MHI_TRE_GET_EV_LEN_MAX_TRB(max_trb, tre) (GENMASK(__fls(max_trb), 0) & \ > + (MHI_TRE_GET_DWORD(tre, 0))) > #define MHI_TRE_GET_EV_CHID(tre) FIELD_GET(GENMASK(31, 24), (MHI_TRE_GET_DWORD(tre, 1))) > #define MHI_TRE_GET_EV_TYPE(tre) FIELD_GET(GENMASK(23, 16), (MHI_TRE_GET_DWORD(tre, 1))) > #define MHI_TRE_GET_EV_STATE(tre) FIELD_GET(GENMASK(31, 24), (MHI_TRE_GET_DWORD(tre, 0))) > @@ -188,6 +190,7 @@ > /* Transfer descriptor macros */ > #define MHI_TRE_DATA_PTR(ptr) cpu_to_le64(ptr) > #define MHI_TRE_DATA_DWORD0(len) cpu_to_le32(FIELD_PREP(GENMASK(15, 0), len)) > +#define MHI_TRE_DATA_DWORD0_MAX_TREB_CAP(max_len, len) ((GENMASK(__fls(max_len), 0)) & (len)) > #define MHI_TRE_TYPE_TRANSFER 2 > #define MHI_TRE_DATA_DWORD1(bei, ieot, ieob, chain) cpu_to_le32(FIELD_PREP(GENMASK(23, 16), \ > MHI_TRE_TYPE_TRANSFER) | \ > @@ -206,7 +209,11 @@ > #define MHI_RSCTRE_DATA_PTR(ptr, len) cpu_to_le64(FIELD_PREP(GENMASK(64, 48), len) | ptr) > #define MHI_RSCTRE_DATA_DWORD0(cookie) cpu_to_le32(cookie) > #define MHI_RSCTRE_DATA_DWORD1 cpu_to_le32(FIELD_PREP(GENMASK(23, 16), \ > - MHI_PKT_TYPE_COALESCING)) > + MHI_PKT_TYPE_COALESCING)) > + > +/* MHI Max TRB Length CAP ID */ > +#define MAX_TRB_LEN_CAP_ID 0x5 > +#define MAX_TRB_LEN_CFG 0x4 > > enum mhi_capability_type { > MHI_CAP_ID_INTX = 0x1, > diff --git a/drivers/bus/mhi/host/init.c b/drivers/bus/mhi/host/init.c > index 50f96f2c823f..b0982cb24fb9 100644 > --- a/drivers/bus/mhi/host/init.c > +++ b/drivers/bus/mhi/host/init.c > @@ -500,6 +500,25 @@ static int mhi_find_capability(struct mhi_controller *mhi_cntrl, u32 capability, > return -ENXIO; > } > > +static int mhi_init_ext_trb_len(struct mhi_controller *mhi_cntrl) Why _ext_? > +{ > + struct device *dev = &mhi_cntrl->mhi_dev->dev; > + u32 trb_offset; > + int ret; > + > + ret = mhi_find_capability(mhi_cntrl, MAX_TRB_LEN_CAP_ID, &trb_offset); > + if (ret) > + return ret; > + > + /* Get max TRB length */ > + ret = mhi_read_reg(mhi_cntrl, mhi_cntrl->regs, > + trb_offset + MAX_TRB_LEN_CFG, &mhi_cntrl->ext_trb_len); > + > + dev_dbg(dev, "Max TRB length supported is : 0x%x\n", mhi_cntrl->ext_trb_len); "Max TRB length: 0x%x" But it feels more natural to expose the value in decimal, than in hex. - Mani -- மணிவண்ணன் சதாசிவம்