openwrt/target/linux/bcm27xx/patches-5.15/950-0914-mmc-block-Don-t-do-single-sector-reads-during-recove.patch
John Audia 1eebe72a80 kernel: bump 5.15 to 5.15.77
Manually rebased:
   bcm27xx/patches-5.15/950-0600-xhci-quirks-add-link-TRB-quirk-for-VL805.patch
   bcm27xx/patches-5.15/950-0606-usb-xhci-add-VLI_TRB_CACHE_BUG-quirk.patch
   bcm27xx/patches-5.15/950-0717-usb-xhci-add-a-quirk-for-Superspeed-bulk-OUT-transfe.patch
   bcm53xx/patches-5.15/180-usb-xhci-add-support-for-performing-fake-doorbell.patch
   lantiq/patches-5.15/0028-NET-lantiq-various-etop-fixes.patch

All other patches automatically rebased

Build system: x86_64
Build-tested: bcm2711/RPi4B, mt7622/RT3200
Run-tested: bcm2711/RPi4B, mt7622/RT3200

Signed-off-by: John Audia <therealgraysky@proton.me>
2022-11-05 16:26:38 +01:00

39 lines
1.4 KiB
Diff

From b70da899940777d92338d8f83f7a6236818ad08a Mon Sep 17 00:00:00 2001
From: Jonathan Bell <jonathan@raspberrypi.com>
Date: Mon, 16 May 2022 10:28:27 +0100
Subject: [PATCH] mmc: block: Don't do single-sector reads during
recovery
See https://github.com/raspberrypi/linux/issues/5019
If an SD card has degraded performance such that IO operations time out
then the MMC block layer will leak SG DMA mappings in the swiotlb during
recovery. It retries the same SG and this causes the leak, as it is
mapped twice - once in sdhci_pre_req() and again during single-block
reads in sdhci_prepare_data().
Resetting the card (including power-cycling if a regulator for vmmc is
present) ought to be enough to recover a stuck state, so for now don't
try single-block reads in the recovery path.
Signed-off-by: Jonathan Bell <jonathan@raspberrypi.com>
---
drivers/mmc/core/block.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
--- a/drivers/mmc/core/block.c
+++ b/drivers/mmc/core/block.c
@@ -1883,7 +1883,11 @@ static void mmc_blk_mq_rw_recovery(struc
return;
}
- if (rq_data_dir(req) == READ && brq->data.blocks >
+ /*
+ * XXX: don't do single-sector reads, as it leaks a SG DMA
+ * mapping when reusing the still-pending req.
+ */
+ if (0 && rq_data_dir(req) == READ && brq->data.blocks >
queue_physical_block_size(mq->queue) >> 9) {
/* Read one (native) sector at a time */
mmc_blk_read_single(mq, req);