linux.git/drivers/tty/serial/8250, branch v5.14

serial: 8250_pci: Avoid irq sharing for MSI(-X) interrupts.

2021-07-30T11:06:19+00:00

This attempts to fix a bug found with a serial port card which uses
an MCS9922 chip, one of the 4 models for which MSI-X interrupts are
currently supported. I don't possess such a card, and i'm not
experienced with the serial subsystem, so this patch is based on what
i think i found as a likely reason for failure, based on walking the
user who actually owns the card through some diagnostic.

The user who reported the problem finds the following in his dmesg
output for the relevant ttyS4 and ttyS5:

[    0.580425] serial 0000:02:00.0: enabling device (0000 -> 0003)
[    0.601448] 0000:02:00.0: ttyS4 at I/O 0x3010 (irq = 125, base_baud = 115200) is a ST16650V2
[    0.603089] serial 0000:02:00.1: enabling device (0000 -> 0003)
[    0.624119] 0000:02:00.1: ttyS5 at I/O 0x3000 (irq = 126, base_baud = 115200) is a ST16650V2
...
[    6.323784] genirq: Flags mismatch irq 128. 00000080 (ttyS5) vs. 00000000 (xhci_hcd)
[    6.324128] genirq: Flags mismatch irq 128. 00000080 (ttyS5) vs. 00000000 (xhci_hcd)
...

Output of setserial -a:

/dev/ttyS4, Line 4, UART: 16650V2, Port: 0x3010, IRQ: 127
	Baud_base: 115200, close_delay: 50, divisor: 0
	closing_wait: 3000
	Flags: spd_normal skip_test

This suggests to me that the serial driver wants to register and share a
MSI/MSI-X irq 128 with the xhci_hcd driver, whereas the xhci driver does
not want to share the irq, as flags 0x00000080 (== IRQF_SHARED) from the
serial port driver means to share the irq, and this mismatch ends in some
failed irq init?

With this setup, data reception works very unreliable, with dropped data,
already at a transmission rate of only a 16 Bytes chunk every 1/120th of
a second, ie. 1920 Bytes/sec, presumably due to rx fifo overflow due to
mishandled or not used at all rx irq's?

See full discussion thread with attempted diagnosis at:

https://psychtoolbox.discourse.group/t/issues-with-iscan-serial-port-recording/3886

Disabling the use of MSI interrupts for the serial port pci card did
fix the reliability problems. The user executed the following sequence
of commands to achieve this:

echo 0000:02:00.0 | sudo tee /sys/bus/pci/drivers/serial/unbind
echo 0000:02:00.1 | sudo tee /sys/bus/pci/drivers/serial/unbind

echo 0 | sudo tee /sys/bus/pci/devices/0000:02:00.0/msi_bus
echo 0 | sudo tee /sys/bus/pci/devices/0000:02:00.1/msi_bus

echo 0000:02:00.0 | sudo tee /sys/bus/pci/drivers/serial/bind
echo 0000:02:00.1 | sudo tee /sys/bus/pci/drivers/serial/bind

This resulted in the following log output:

[   82.179021] pci 0000:02:00.0: MSI/MSI-X disallowed for future drivers
[   87.003031] pci 0000:02:00.1: MSI/MSI-X disallowed for future drivers
[   98.537010] 0000:02:00.0: ttyS4 at I/O 0x3010 (irq = 17, base_baud = 115200) is a ST16650V2
[  103.648124] 0000:02:00.1: ttyS5 at I/O 0x3000 (irq = 18, base_baud = 115200) is a ST16650V2

This patch attempts to fix the problem by disabling irq sharing when
using MSI irq's. Note that all i know for sure is that disabling MSI
irq's fixed the problem for the user, so this patch could be wrong and
is untested. Please review with caution, keeping this in mind.

Fixes: 8428413b1d14 ("serial: 8250_pci: Implement MSI(-X) support")
Cc: Ralf Ramsauer 
Cc: stable 
Reviewed-by: Andy Shevchenko 
Signed-off-by: Mario Kleiner 
Link: https://lore.kernel.org/r/20210729043306.18528-1-mario.kleiner.de@gmail.com
Signed-off-by: Greg Kroah-Hartman

serial: 8250_mtk: fix uart corruption issue when rx power off

2021-07-29T15:06:37+00:00

Fix uart corruption issue when rx power off.
Add spin lock in mtk8250_dma_rx_complete function in APDMA mode.

when uart is used as a communication port with external device(GPS).
when external device(GPS) power off, the power of rx pin is also from
1.8v to 0v. Even if there is not any data in rx. But uart rx pin can
capture the data "0".
If uart don't receive any data in specified cycle, uart will generates
BI(Break interrupt) interrupt.
If external device(GPS) power off, we found that BI interrupt appeared
continuously and very frequently.
When uart interrupt type is BI, uart IRQ handler(8250 framwork
API:serial8250_handle_irq) will push data to tty buffer.
mtk8250_dma_rx_complete is a task of mtk_uart_apdma_rx_handler.
mtk8250_dma_rx_complete priority is lower than uart irq
handler(serial8250_handle_irq).
if we are in process of mtk8250_dma_rx_complete, uart appear BI
interrupt:1)serial8250_handle_irq will priority execution.2)it may cause
write tty buffer conflict in mtk8250_dma_rx_complete.
So the spin lock protect the rx receive data process is not break.

Signed-off-by: Zhiyong Tao 
Cc: stable 
Link: https://lore.kernel.org/r/20210729084640.17613-2-zhiyong.tao@mediatek.com
Signed-off-by: Greg Kroah-Hartman

serial: 8250_pci: Enumerate Elkhart Lake UARTs via dedicated driver

2021-07-21T10:54:11+00:00

Elkhart Lake UARTs are PCI enumerated Synopsys DesignWare v4.0+ UART
integrated with Intel iDMA 32-bit DMA controller. There is a specific
driver to handle them, i.e. 8250_lpss. Hence, disable 8250_pci
enumeration for these UARTs.

Fixes: 1b91d97c66ef ("serial: 8250_lpss: Add ->setup() for Elkhart Lake ports")
Fixes: 4f912b898dc2 ("serial: 8250_lpss: Enable HS UART on Elkhart Lake")
Cc: stable 
Signed-off-by: Andy Shevchenko 
Link: https://lore.kernel.org/r/20210713101739.36962-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Greg Kroah-Hartman

serial: 8250: fix handle_irq locking

2021-07-21T10:53:26+00:00

The 8250 handle_irq callback is not just called from the interrupt
handler but also from a timer callback when polling (e.g. for ports
without an interrupt line). Consequently the callback must explicitly
disable interrupts to avoid a potential deadlock with another interrupt
in polled mode.

Add back an irqrestore-version of the sysrq port-unlock helper and use
it in the 8250 callbacks that need it.

Fixes: 75f4e830fa9c ("serial: do not restore interrupt state in sysrq helper")
Cc: stable@vger.kernel.org	# 5.13
Cc: Joel Stanley 
Cc: Andrew Jeffery 
Reported-by: kernel test robot 
Signed-off-by: Johan Hovold 
Link: https://lore.kernel.org/r/20210714080427.28164-1-johan@kernel.org
Signed-off-by: Greg Kroah-Hartman

serial: 8250: Mask out floating 16/32-bit bus bits

2021-07-21T10:52:14+00:00

Make sure only actual 8 bits of the IIR register are used in determining
the port type in `autoconfig'.

The `serial_in' port accessor returns the `unsigned int' type, meaning
that with UPIO_AU, UPIO_MEM16, UPIO_MEM32, and UPIO_MEM32BE access types
more than 8 bits of data are returned, of which the high order bits will
often come from bus lines that are left floating in the data phase.  For
example with the MIPS Malta board's CBUS UART, where the registers are
aligned on 8-byte boundaries and which uses 32-bit accesses, data as
follows is returned:

YAMON> dump -32 0xbf000900 0x40

BF000900: 1F000942 1F000942 1F000900 1F000900  ...B...B........
BF000910: 1F000901 1F000901 1F000900 1F000900  ................
BF000920: 1F000900 1F000900 1F000960 1F000960  ...........`...`
BF000930: 1F000900 1F000900 1F0009FF 1F0009FF  ................

YAMON>

Evidently high-order 24 bits return values previously driven in the
address phase (the 3 highest order address bits used with the command
above are masked out in the simple virtual address mapping used here and
come out at zeros on the external bus), a common scenario with bus lines
left floating, due to bus capacitance.

Consequently when the value of IIR, mapped at 0x1f000910, is retrieved
in `autoconfig', it comes out at 0x1f0009c1 and when it is right-shifted
by 6 and then assigned to 8-bit `scratch' variable, the value calculated
is 0x27, not one of 0, 1, 2, 3 expected in port type determination.

Fix the issue then, by assigning the value returned from `serial_in' to
`scratch' first, which masks out 24 high-order bits retrieved, and only
then right-shift the resulting 8-bit data quantity, producing the value
of 3 in this case, as expected.  Fix the same issue in `serial_dl_read'.

The problem first appeared with Linux 2.6.9-rc3 which predates our repo
history, but the origin could be identified with the old MIPS/Linux repo
also at: 
as commit e0d2356c0777 ("Merge with Linux 2.6.9-rc3."), where code in
`serial_in' was updated with this case:

+	case UPIO_MEM32:
+		return readl(up->port.membase + offset);
+

which made it produce results outside the unsigned 8-bit range for the
first time, though obviously it is system dependent what actual values
appear in the high order bits retrieved and it may well have been zeros
in the relevant positions with the system the change originally was
intended for.  It is at that point that code in `autoconf' should have
been updated accordingly, but clearly it was overlooked.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Cc: stable@vger.kernel.org # v2.6.12+
Reviewed-by: Philippe Mathieu-Daudé 
Signed-off-by: Maciej W. Rozycki 
Link: https://lore.kernel.org/r/alpine.DEB.2.21.2106260516220.37803@angie.orcam.me.uk
Signed-off-by: Greg Kroah-Hartman

serial: 8250: 8250_omap: Fix possible interrupt storm on K3 SoCs

2021-06-24T12:51:33+00:00

On K3 family of SoCs (which includes AM654 SoC), it is observed that RX
TIMEOUT is signalled after RX FIFO has been drained, in which case a
dummy read of RX FIFO is required to clear RX TIMEOUT condition.
Otherwise, this would lead to an interrupt storm.

Fix this by introducing UART_RX_TIMEOUT_QUIRK flag and doing a dummy
read in IRQ handler when RX TIMEOUT is reported with no data in RX FIFO.

Fixes: be70874498f3 ("serial: 8250_omap: Add support for AM654 UART controller")
Reported-by: Jan Kiszka 
Tested-by: Jan Kiszka 
Signed-off-by: Vignesh Raghavendra 
Link: https://lore.kernel.org/r/20210622145704.11168-1-vigneshr@ti.com
Signed-off-by: Greg Kroah-Hartman

serial_cs: Add Option International GSM-Ready 56K/ISDN modem

2021-06-16T07:20:29+00:00

Add support for Option International GSM-Ready 56K/ISDN PCMCIA modem
card.

Signed-off-by: Ondrej Zary 
Cc: stable 
Link: https://lore.kernel.org/r/20210611201940.23898-2-linux@zary.sk
Signed-off-by: Greg Kroah-Hartman

serial_cs: remove wrong GLOBETROTTER.cis entry

2021-06-16T07:20:29+00:00

The GLOBETROTTER.cis entry in serial_cs matches more devices than
intended and breaks them. Remove it.

Example: # pccardctl info
PRODID_1="Option International
"
PRODID_2="GSM-Ready 56K/ISDN
"
PRODID_3="021
"
PRODID_4="A
"
MANFID=0013,0000
FUNCID=0

result:
pcmcia 0.0: Direct firmware load for cis/GLOBETROTTER.cis failed with error -2

The GLOBETROTTER.cis is nowhere to be found. There's GLOBETROTTER.cis.ihex at
https://netdev.vger.kernel.narkive.com/h4inqdxM/patch-axnet-cs-fix-phy-id-detection-for-bogus-asix-chip#post41
It's from completely diffetent card:
vers_1 4.1, "Option International", "GSM/GPRS GlobeTrotter", "001", "A"

Signed-off-by: Ondrej Zary 
Cc: stable 
Link: https://lore.kernel.org/r/20210611201940.23898-1-linux@zary.sk
Signed-off-by: Greg Kroah-Hartman

serial: 8250: Handle custom baud rates in UPF_MAGIC_MULTIPLIER range

2021-06-16T07:20:29+00:00

Handle custom baud rates requested in the UPF_MAGIC_MULTIPLIER range
with BOTHER.  Currently matching is exact, that is if a baud rate that
is not either of clk/4 or clk/8 is requested, then we fall through to
the default case, which will just divide the clock rate by 16 times the
rate requested, round it to closest integer, and possibly yield even
worse results then if clamping to the extra baud rates was chosen.

So for example if we have the usual base rate of 115200 and request a
rate of 230399, then the fall-through divisor calculation will yield 1,
and consequently the baud rate of 115200 will be programmed even though
obviously the magic rate of 230400 would be more appropriate.

Make the selection of the magic rates range-qualified then and use clk/4
for rates from clk/6 up (assuming `serial8250_get_baud_rate' has already
rejected any rates too far beyond clk/4), and otherwise use clk/8 for
rates from clk/12 up, and finally fall through to the standard divisor
calculation.  Explicitly void using the undocumented rate of 153600bps
and stick to documented divisor values only.

Signed-off-by: Maciej W. Rozycki 
Link: https://lore.kernel.org/r/alpine.DEB.2.21.2105200232090.29169@angie.orcam.me.uk
Signed-off-by: Greg Kroah-Hartman

serial: 8250: Actually allow UPF_MAGIC_MULTIPLIER baud rates

2021-06-16T07:20:29+00:00

Support for magic baud rate divisors of 32770 and 32769 used with SMSC
Super I/O chips for extra baud rates of 230400 and 460800 respectively
where base rate is 115200[1] has been added around Linux 2.5.64, which
predates our repo history, but the origin could be identified as commit
2a717aad772f ("Merge with Linux 2.5.64.") with the old MIPS/Linux repo
also at: .

Code that is now in `serial8250_do_get_divisor' was added back then to
`serial8250_get_divisor', but that code would only ever trigger if one
of the higher baud rates was actually requested, and that cannot ever
happen, because the earlier call to `serial8250_get_baud_rate' never
returns them.  This is because it calls `uart_get_baud_rate' with the
maximum requested being the base rate, that is clk/16 or 115200 for SMSC
chips at their nominal clock rate.

Fix it then and allow UPF_MAGIC_MULTIPLIER baud rates to be selected, by
requesting the maximum baud rate of clk/4 rather than clk/16 if the flag
has been set.  Also correct the minimum baud rate, observing that these
ports only support actual (non-magic) divisors of up to 32767 only.

References:

[1] "FDC37M81x, PC98/99 Compliant Enhanced Super I/O Controller with
    Keyboard/Mouse Wake-Up", Standard Microsystems Corporation, Rev.
    03/27/2000, Table 31 - "Baud Rates", p. 77

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Maciej W. Rozycki 
Link: https://lore.kernel.org/r/alpine.DEB.2.21.2105190412280.29169@angie.orcam.me.uk
Signed-off-by: Greg Kroah-Hartman