linux-kernel - [PATCH v6 5/6] dmaengine: dw-edma: HDMA: Add sync read before starting the DMA transfer in remote setup

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: <20231117-b4-feature_hdma_mainline-v6-5-ebf7aa0e40d7@bootlin.com>
Date:   Fri, 17 Nov 2023 11:03:53 +0100
From:   Kory Maincent <kory.maincent@...tlin.com>
To:     Manivannan Sadhasivam <mani@...nel.org>,
        Gustavo Pimentel <gustavo.pimentel@...opsys.com>,
        Serge Semin <fancer.lancer@...il.com>,
        Vinod Koul <vkoul@...nel.org>,
        Cai Huoqing <cai.huoqing@...ux.dev>
Cc:     Thomas Petazzoni <thomas.petazzoni@...tlin.com>,
        dmaengine@...r.kernel.org, linux-kernel@...r.kernel.org,
        Herve Codina <herve.codina@...tlin.com>,
        Kory Maincent <kory.maincent@...tlin.com>,
        Manivannan Sadhasivam <mani@...nel.org>
Subject: [PATCH v6 5/6] dmaengine: dw-edma: HDMA: Add sync read before
 starting the DMA transfer in remote setup

The Linked list element and pointer are not stored in the same memory as
the HDMA controller register. If the doorbell register is toggled before
the full write of the linked list a race condition error will occur.
In remote setup we can only use a readl to the memory to assure the full
write has occurred.

Fixes: e74c39573d35 ("dmaengine: dw-edma: Add support for native HDMA")
Reviewed-by: Serge Semin <fancer.lancer@...il.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@...aro.org>
Signed-off-by: Kory Maincent <kory.maincent@...tlin.com>
---

Changes in v2:
- Move the sync read in a function.
- Add commments

Changes in v4:
- Update git commit message.

Changes in v6:
- Fix comment typos.
---
 drivers/dma/dw-edma/dw-hdma-v0-core.c | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/drivers/dma/dw-edma/dw-hdma-v0-core.c b/drivers/dma/dw-edma/dw-hdma-v0-core.c
index 04b0bcb6ded9..10e8f0715114 100644
--- a/drivers/dma/dw-edma/dw-hdma-v0-core.c
+++ b/drivers/dma/dw-edma/dw-hdma-v0-core.c
@@ -222,6 +222,20 @@ static void dw_hdma_v0_core_write_chunk(struct dw_edma_chunk *chunk)
 	dw_hdma_v0_write_ll_link(chunk, i, control, chunk->ll_region.paddr);
 }
 
+static void dw_hdma_v0_sync_ll_data(struct dw_edma_chunk *chunk)
+{
+	/*
+	 * In case of remote HDMA engine setup, the DW PCIe RP/EP internal
+	 * configuration registers and application memory are normally accessed
+	 * over different buses. Ensure LL-data reaches the memory before the
+	 * doorbell register is toggled by issuing the dummy-read from the remote
+	 * LL memory in a hope that the MRd TLP will return only after the
+	 * last MWr TLP is completed
+	 */
+	if (!(chunk->chan->dw->chip->flags & DW_EDMA_CHIP_LOCAL))
+		readl(chunk->ll_region.vaddr.io);
+}
+
 static void dw_hdma_v0_core_start(struct dw_edma_chunk *chunk, bool first)
 {
 	struct dw_edma_chan *chan = chunk->chan;
@@ -252,6 +266,9 @@ static void dw_hdma_v0_core_start(struct dw_edma_chunk *chunk, bool first)
 	/* Set consumer cycle */
 	SET_CH_32(dw, chan->dir, chan->id, cycle_sync,
 		  HDMA_V0_CONSUMER_CYCLE_STAT | HDMA_V0_CONSUMER_CYCLE_BIT);
+
+	dw_hdma_v0_sync_ll_data(chunk);
+
 	/* Doorbell */
 	SET_CH_32(dw, chan->dir, chan->id, doorbell, HDMA_V0_DOORBELL_START);
 }

-- 
2.25.1