linux-kernel - [PATCH 6/9] mmc: sdhci: allow drivers to pre-allocate bounce buffer

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20250923-v4-patch-final-v1-6-2283ad7cbf88@thundersoft.com>
Date: Tue, 23 Sep 2025 14:10:12 +0800
From: Albert Yang <yangzh0906@...ndersoft.com>
To: Rob Herring <robh@...nel.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>, 
 Conor Dooley <conor+dt@...nel.org>, Ge Gordon <gordon.ge@....ai>, 
 BST Linux Kernel Upstream Group <bst-upstream@...ai.top>, 
 Catalin Marinas <catalin.marinas@....com>, Will Deacon <will@...nel.org>, 
 Ulf Hansson <ulf.hansson@...aro.org>, 
 Adrian Hunter <adrian.hunter@...el.com>, Arnd Bergmann <arnd@...db.de>
Cc: devicetree@...r.kernel.org, linux-kernel@...r.kernel.org, 
 linux-arm-kernel@...ts.infradead.org, linux-mmc@...r.kernel.org, 
 soc@...ts.linux.dev, Albert Yang <yangzh0906@...ndersoft.com>
Subject: [PATCH 6/9] mmc: sdhci: allow drivers to pre-allocate bounce
 buffer

In sdhci_allocate_bounce_buffer(), add an early path that respects a
driver-provided pre-allocated bounce buffer (host->bounce_buffer).

If the buffer is already allocated by the driver (e.g. coherent/SRAM
buffer needed for platforms with 32-bit DMA constraints), just compute
max_blocks from host->bounce_buffer_size and jump to the common "out"
path to set mmc->max_*.

This enables platform drivers to allocate the bounce buffer before
sdhci_add_host(), avoiding starting the host without the buffer ready
and aligning with the guidance from review.

No functional change for drivers that do not pre-allocate the buffer.

drivers/mmc/host/sdhci.c (sdhci_allocate_bounce_buffer): Handle
pre-allocated bounce buffer and fall through to set mmc->max_*.

Suggested-by: Adrian Hunter <adrian.hunter@...el.com>
Link: https://lore.kernel.org/lkml/2b23bcb9-abc7-4667-b939-a19ecae935a2@intel.com/
Signed-off-by: Albert Yang <yangzh0906@...ndersoft.com>

---
Changes for v4:
- Add new patch by Suggested-by: Adrian Hunter
---
 drivers/mmc/host/sdhci.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/mmc/host/sdhci.c b/drivers/mmc/host/sdhci.c
index 3a17821efa5ca92c6c29141d8fcb9ebf58355cc7..03fbe30cb205e16b924481caa44f0979d230f380 100644
--- a/drivers/mmc/host/sdhci.c
+++ b/drivers/mmc/host/sdhci.c
@@ -4193,6 +4193,12 @@ static void sdhci_allocate_bounce_buffer(struct sdhci_host *host)
 	unsigned int bounce_size;
 	int ret;
 
+	/* Drivers may have already allocated the buffer */
+	if (host->bounce_buffer) {
+		bounce_size = host->bounce_buffer_size;
+		max_blocks = bounce_size / 512;
+		goto out;
+	}
 	/*
 	 * Cap the bounce buffer at 64KB. Using a bigger bounce buffer
 	 * has diminishing returns, this is probably because SD/MMC
@@ -4241,6 +4247,7 @@ static void sdhci_allocate_bounce_buffer(struct sdhci_host *host)
 
 	host->bounce_buffer_size = bounce_size;
 
+out:
 	/* Lie about this since we're bouncing */
 	mmc->max_segs = max_blocks;
 	mmc->max_seg_size = bounce_size;

-- 
2.43.0