| Message ID | c53d287bf741831e1eb19137091c00efc3f9b3c8.1662124370.git-series.marmarek@invisiblethingslab.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Add Xue - console over USB 3 Debug Capability | expand |
On 02.09.2022 15:17, Marek Marczykowski-Górecki wrote: > When cable is unplugged, dbc_ensure_running() correctly detects this > situation (DBC_CTRL_DCR flag is clear), and prevent sending data > immediately to the device. It gets only queued in work ring buffers. > When cable is plugged in again, subsequent dbc_flush() will send the > buffered data. > But there is a corner case, where no subsequent data was buffered in the > work buffer, but a TRB was still pending. Ring the doorbell to let the > controller re-send them. For console output it is rare corner case (TRB > is pending for a very short time), but for console input it is very > normal case (there is always one pending TRB for input). > > Extract doorbell ringing into separate function to avoid duplication. > > Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com> Acked-by: Jan Beulich <jbeulich@suse.com>
diff --git a/xen/drivers/char/xhci-dbc.c b/xen/drivers/char/xhci-dbc.c index 9f7e1dd60a78..829f1d1d910f 100644 --- a/xen/drivers/char/xhci-dbc.c +++ b/xen/drivers/char/xhci-dbc.c @@ -554,6 +554,15 @@ static unsigned int dbc_work_ring_space_to_end(const struct dbc_work_ring *ring) return ring->deq - ring->enq; } +static void dbc_ring_doorbell(struct dbc *dbc, int doorbell) +{ + uint32_t __iomem *db_reg = &dbc->dbc_reg->db; + uint32_t db = (readl(db_reg) & ~DBC_DOORBELL_TARGET_MASK) | + (doorbell << DBC_DOORBELL_TARGET_SHIFT); + + writel(db, db_reg); +} + static void dbc_push_trb(struct dbc *dbc, struct xhci_trb_ring *ring, uint64_t dma, uint64_t len) { @@ -1024,6 +1033,8 @@ static bool dbc_ensure_running(struct dbc *dbc) writel(ctrl | (1U << DBC_CTRL_DRC), ®->ctrl); writel(readl(®->portsc) | (1U << DBC_PSC_PED), ®->portsc); wmb(); + dbc_ring_doorbell(dbc, dbc->dbc_iring.db); + dbc_ring_doorbell(dbc, dbc->dbc_oring.db); } return true; @@ -1041,10 +1052,6 @@ static bool dbc_ensure_running(struct dbc *dbc) static void dbc_flush(struct dbc *dbc, struct xhci_trb_ring *trb, struct dbc_work_ring *wrk) { - struct dbc_reg *reg = dbc->dbc_reg; - uint32_t db = (readl(®->db) & ~DBC_DOORBELL_TARGET_MASK) | - (trb->db << DBC_DOORBELL_TARGET_SHIFT); - if ( xhci_trb_ring_full(trb) ) return; @@ -1068,7 +1075,7 @@ static void dbc_flush(struct dbc *dbc, struct xhci_trb_ring *trb, } wmb(); - writel(db, ®->db); + dbc_ring_doorbell(dbc, trb->db); } /**
When cable is unplugged, dbc_ensure_running() correctly detects this situation (DBC_CTRL_DCR flag is clear), and prevent sending data immediately to the device. It gets only queued in work ring buffers. When cable is plugged in again, subsequent dbc_flush() will send the buffered data. But there is a corner case, where no subsequent data was buffered in the work buffer, but a TRB was still pending. Ring the doorbell to let the controller re-send them. For console output it is rare corner case (TRB is pending for a very short time), but for console input it is very normal case (there is always one pending TRB for input). Extract doorbell ringing into separate function to avoid duplication. Signed-off-by: Marek Marczykowski-Górecki <marmarek@invisiblethingslab.com> --- Changes in v6: - keep barriers consistent --- xen/drivers/char/xhci-dbc.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-)