openwrt/target/linux/bcm27xx/patches-5.15/950-0061-tty-amba-pl011-Avoid-rare-write-when-full-error.patch

43 lines
1.7 KiB
Diff
Raw Normal View History

From db675d9894cea3a30ba50695a3e1ea9739782bd2 Mon Sep 17 00:00:00 2001
From: Phil Elwell <phil@raspberrypi.org>
Date: Wed, 29 Jan 2020 09:35:19 +0000
Subject: [PATCH] tty: amba-pl011: Avoid rare write-when-full error
Under some circumstances on BCM283x processors data loss can be
observed - a single byte missing from the TX output stream. These bytes
are always the last byte of a batch of 8 written from pl011_tx_chars
when from_irq is true, meaning that the FIFO full flag is not checked
before writing.
The transmit optimisation relies on the FIFO being half-empty when the
TX interrupt is raised. Instrumenting the driver further showed that
the failure case correlated with the TX FIFO full flag being set at the
point where the last byte was written to the data register, which
explains the data loss but not how the FIFO appeared to be prematurely
full. A possible explanation is that a FIFO write was in flight at the
time the interrupt was raised, but as yet there is no hypothesis as to
how this might occur.
In the absence of a clear understanding of the failure mechanism, avoid
the problem by checking the FIFO levels before writing the last byte of
the group, which will have minimal performance impact.
Signed-off-by: Phil Elwell <phil@raspberrypi.org>
---
drivers/tty/serial/amba-pl011.c | 4 ++++
1 file changed, 4 insertions(+)
--- a/drivers/tty/serial/amba-pl011.c
+++ b/drivers/tty/serial/amba-pl011.c
kernel: bump 5.15 to 5.15.86 Removed upstreamed: pending-5.15/101-Use-stddefs.h-instead-of-compiler.h.patch[1] ipq806x/patches-5.15/122-01-clk-qcom-clk-krait-fix-wrong-div2-functions.patch[2] bcm27xx/patches-5.15/950-0198-drm-fourcc-Add-packed-10bit-YUV-4-2-0-format.patch[3] Manually rebased: ramips/patches-5.15/100-PCI-mt7621-Add-MediaTek-MT7621-PCIe-host-controller-.patch[4] Added patch/backported: ramips/patches-5.15/107-PCI-mt7621-Add-sentinel-to-quirks-table.patch[5] All other patches automatically rebased. 1. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v5.15.86&id=c160505c9b574b346031fdf2c649d19e7939ca11 2. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v5.15.86&id=a051e10bfc6906d29dae7a31f0773f2702edfe1b 3. https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v5.15.86&id=ec1727f89ecd6f2252c0c75e200058819f7ce47a 4. Quilt gave this output when I applied the patch to rebase it: % quilt push -f Applying patch platform/100-PCI-mt7621-Add-MediaTek-MT7621-PCIe-host-controller-.patch patching file arch/mips/ralink/Kconfig patching file drivers/pci/controller/Kconfig patching file drivers/pci/controller/Makefile patching file drivers/staging/Kconfig patching file drivers/staging/Makefile patching file drivers/staging/mt7621-pci/Kconfig patching file drivers/staging/mt7621-pci/Makefile patching file drivers/staging/mt7621-pci/TODO patching file drivers/staging/mt7621-pci/mediatek,mt7621-pci.txt patching file drivers/staging/mt7621-pci/pci-mt7621.c Hunk #1 FAILED at 1. Not deleting file drivers/staging/mt7621-pci/pci-mt7621.c as content differs from patch 1 out of 1 hunk FAILED -- saving rejects to file drivers/staging/mt7621-pci/pci-mt7621.c.rej patching file drivers/pci/controller/pcie-mt7621.c Applied patch platform/100-PCI-mt7621-Add-MediaTek-MT7621-PCIe-host-controller-.patch (forced; needs refresh) Upon inspecting drivers/staging/mt7621-pci/pci-mt7621.c.rej, it seems that the original patch wants to delete drivers/staging/mt7621-pci/pci-mt7621.c but upstream's version was not an exact match. I opted to delete that file. 5. Suggestion by hauke: https://github.com/torvalds/linux/commit/19098934f910b4d47cb30251dd39ffa57bef9523 "This patch is in upstream kernel, but it was backported to the old staging driver in kernel 5.15." Build system: x86_64 Build-tested: bcm2711/RPi4B, filogic/xiaomi_redmi-router-ax6000-ubootmod Run-tested: bcm2711/RPi4B, filogic/xiaomi_redmi-router-ax6000-ubootmod Signed-off-by: John Audia <therealgraysky@proton.me>
2022-12-31 15:29:53 +00:00
@@ -1496,6 +1496,10 @@ static bool pl011_tx_chars(struct uart_a
if (likely(from_irq) && count-- == 0)
break;
+ if (likely(from_irq) && count == 0 &&
+ pl011_read(uap, REG_FR) & UART01x_FR_TXFF)
+ break;
+
if (!pl011_tx_char(uap, xmit->buf[xmit->tail], from_irq))
break;