Remove the check for zero-length packets as this condition cannot
occur during normal operation. According to the Xilinx AXI4-Stream
FIFO Product Guide (PG080), in the Receive Length Register (RLR)
description: "The smallest packet that can be received is 1 byte."
A zero-length packet would indicate a bug in the IP core itself.
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://patch.msgid.link/20251227212640.3321310-6-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Module parameters for timeouts are a poor interface choice as they
affect all device instances globally rather than being configurable
per file descriptor.
The current implementation also returns -EAGAIN on timeout, requiring
userspace to implement retry loops around blocking operations.
Remove the read_timeout and write_timeout module parameters. The next
commit adds poll() support, allowing applications to implement timeout
handling using standard poll()/select() interfaces.
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://patch.msgid.link/20251227212640.3321310-1-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The irq handler iterates through the interrupt bits one by one in a while()
loop, clearing one bit at a time. This is unnecessary and inefficient, as
the 'Interrupt Status Register' can be cleared with a single write
operation.
Also the irq handler checks for some interrupts that are masked out in the
'Interrupt Enable Register', so they are never triggered.
Simplify the irq handler by removing the checks for interrupts that are
masked out and clear the whole 'Interrupt Status Register' in one go. The
macros that are no longer relevant are also removed.
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://lore.kernel.org/r/20250919195400.3180039-2-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Pull staging driver fixes from Greg KH:
"Here are some staging driver fixes that missed 6.17-final due to my
travel schedule. They fix a number of reported issues in the axis-fifo
driver, one of which was just independently discovered by someone else
today so someone is looking at this code.
All of these fixes have been in linux-next for many weeks with no
reported issues"
* tag 'staging-6.18-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: axis-fifo: flush RX FIFO on read errors
staging: axis-fifo: fix TX handling on copy_from_user() failure
staging: axis-fifo: fix maximum TX packet length check
If copy_from_user() fails, write() currently returns -EFAULT, but any
partially written data leaves the TX FIFO in an inconsistent state.
Subsequent write() calls then fail with "transmit length mismatch"
errors.
Once partial data is written to the hardware FIFO, it cannot be removed
without a TX reset. Commit c6e8d85faf ("staging: axis-fifo: Remove
hardware resets for user errors") removed a full FIFO reset for this case,
which fixed a potential RX data loss, but introduced this TX issue.
Fix this by introducing a bounce buffer: copy the full packet from
userspace first, and write to the hardware FIFO only if the copy
was successful.
Fixes: c6e8d85faf ("staging: axis-fifo: Remove hardware resets for user errors")
Cc: stable@vger.kernel.org
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://lore.kernel.org/r/20250912101322.1282507-1-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since commit 2ca34b5087 ("staging: axis-fifo: Correct handling of
tx_fifo_depth for size validation"), write() operations with packets
larger than 'tx_fifo_depth - 4' words are no longer rejected with -EINVAL.
Fortunately, the packets are not actually getting transmitted to hardware,
otherwise they would be raising a 'Transmit Packet Overrun Error'
interrupt, which requires a reset of the TX circuit to recover from.
Instead, the request times out inside wait_event_interruptible_timeout()
and always returns -EAGAIN, since the wake up condition can never be true
for these packets. But still, they unnecessarily block other tasks from
writing to the FIFO and the EAGAIN return code signals userspace to retry
the write() call, even though it will always fail and time out.
According to the AXI4-Stream FIFO reference manual (PG080), the maximum
valid packet length is 'tx_fifo_depth - 4' words, so attempting to send
larger packets is invalid and should not be happening in the first place:
> The maximum packet that can be transmitted is limited by the size of
> the FIFO, which is (C_TX_FIFO_DEPTH–4)*(data interface width/8) bytes.
Therefore, bring back the old behavior and outright reject packets larger
than 'tx_fifo_depth - 4' with -EINVAL. Add a comment to explain why the
check is necessary. The dev_err() message was removed to avoid cluttering
the dmesg log if an invalid packet is received from userspace.
Fixes: 2ca34b5087 ("staging: axis-fifo: Correct handling of tx_fifo_depth for size validation")
Cc: stable@vger.kernel.org
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://lore.kernel.org/r/20250817171350.872105-1-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Axis-fifo devices use physical addresses in their name, for example
'axis_fifo_0x43c00000'. This is generally frowned upon and it does not
follow the usual naming scheme that uses unique identifiers.
Therefore, use ida_alloc()/ida_free() to implement unique identifiers
for axis-fifo device names (i.e. axis_fifo0, axis_fifo1, etc.).
Signed-off-by: Ovidiu Panait <ovidiu.panait.oss@gmail.com>
Link: https://lore.kernel.org/r/20250811085417.2641674-2-ovidiu.panait.oss@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
