mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Oded Gabbay <ogabbay@kernel.org>
To: linux-kernel@vger.kernel.org
Cc: Ohad Sharabi <osharabi@habana.ai>
Subject: [PATCH 10/12] habanalabs: skip read fw errors if dynamic descriptor invalid
Date: Sun, 28 Nov 2021 21:34:33 +0200	[thread overview]
Message-ID: <20211128193435.266534-10-ogabbay@kernel.org> (raw)
In-Reply-To: <20211128193435.266534-1-ogabbay@kernel.org>

From: Ohad Sharabi <osharabi@habana.ai>

Reporting FW errors involves reading of the error registers.

In case we have a corrupted FW descriptor we cannot do that since the
dynamic scratchpad is potentially corrupted as well and may cause kernel
crush when attempting access to a corrupted register offset.

Signed-off-by: Ohad Sharabi <osharabi@habana.ai>
Reviewed-by: Oded Gabbay <ogabbay@kernel.org>
Signed-off-by: Oded Gabbay <ogabbay@kernel.org>
---
 drivers/misc/habanalabs/common/firmware_if.c | 17 +++++++++++++++--
 drivers/misc/habanalabs/common/habanalabs.h  |  2 ++
 2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/drivers/misc/habanalabs/common/firmware_if.c b/drivers/misc/habanalabs/common/firmware_if.c
index bafadcc65497..760ca139d5cf 100644
--- a/drivers/misc/habanalabs/common/firmware_if.c
+++ b/drivers/misc/habanalabs/common/firmware_if.c
@@ -1772,6 +1772,9 @@ static int hl_fw_dynamic_validate_descriptor(struct hl_device *hdev,
 		return rc;
 	}
 
+	/* here we can mark the descriptor as valid as the content has been validated */
+	fw_loader->dynamic_loader.fw_desc_valid = true;
+
 	return 0;
 }
 
@@ -1828,7 +1831,13 @@ static int hl_fw_dynamic_read_and_validate_descriptor(struct hl_device *hdev,
 		return rc;
 	}
 
-	/* extract address copy the descriptor from */
+	/*
+	 * extract address to copy the descriptor from
+	 * in addition, as the descriptor value is going to be over-ridden by new data- we mark it
+	 * as invalid.
+	 * it will be marked again as valid once validated
+	 */
+	fw_loader->dynamic_loader.fw_desc_valid = false;
 	src = hdev->pcie_bar[region->bar_id] + region->offset_in_bar +
 							response->ram_offset;
 	memcpy_fromio(fw_desc, src, sizeof(struct lkd_fw_comms_desc));
@@ -2317,6 +2326,9 @@ static int hl_fw_dynamic_init_cpu(struct hl_device *hdev,
 	dev_info(hdev->dev,
 		"Loading firmware to device, may take some time...\n");
 
+	/* initialize FW descriptor as invalid */
+	fw_loader->dynamic_loader.fw_desc_valid = false;
+
 	/*
 	 * In this stage, "cpu_dyn_regs" contains only LKD's hard coded values!
 	 * It will be updated from FW after hl_fw_dynamic_request_descriptor().
@@ -2412,7 +2424,8 @@ static int hl_fw_dynamic_init_cpu(struct hl_device *hdev,
 	return 0;
 
 protocol_err:
-	fw_read_errors(hdev, le32_to_cpu(dyn_regs->cpu_boot_err0),
+	if (fw_loader->dynamic_loader.fw_desc_valid)
+		fw_read_errors(hdev, le32_to_cpu(dyn_regs->cpu_boot_err0),
 				le32_to_cpu(dyn_regs->cpu_boot_err1),
 				le32_to_cpu(dyn_regs->cpu_boot_dev_sts0),
 				le32_to_cpu(dyn_regs->cpu_boot_dev_sts1));
diff --git a/drivers/misc/habanalabs/common/habanalabs.h b/drivers/misc/habanalabs/common/habanalabs.h
index c2129c9fe9e4..77ac4bb98137 100644
--- a/drivers/misc/habanalabs/common/habanalabs.h
+++ b/drivers/misc/habanalabs/common/habanalabs.h
@@ -1034,6 +1034,7 @@ struct fw_response {
  * @image_region: region to copy the FW image to
  * @fw_image_size: size of FW image to load
  * @wait_for_bl_timeout: timeout for waiting for boot loader to respond
+ * @fw_desc_valid: true if FW descriptor has been validated and hence the data can be used
  */
 struct dynamic_fw_load_mgr {
 	struct fw_response response;
@@ -1041,6 +1042,7 @@ struct dynamic_fw_load_mgr {
 	struct pci_mem_region *image_region;
 	size_t fw_image_size;
 	u32 wait_for_bl_timeout;
+	bool fw_desc_valid;
 };
 
 /**
-- 
2.25.1


  parent reply	other threads:[~2021-11-28 19:37 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-11-28 19:34 [PATCH 01/12] habanalabs: fix soft reset accounting Oded Gabbay
2021-11-28 19:34 ` [PATCH 02/12] habanalabs: rename late init after reset function Oded Gabbay
2021-11-28 19:34 ` [PATCH 03/12] habanalabs/gaudi: return EPERM on non hard-reset Oded Gabbay
2021-11-28 19:34 ` [PATCH 04/12] habanalabs: move device boot warnings to the correct location Oded Gabbay
2021-11-28 19:34 ` [PATCH 05/12] habanalabs: fix race condition in multi CS completion Oded Gabbay
2021-11-28 19:34 ` [PATCH 06/12] habanalabs: add more info ioctls support during reset Oded Gabbay
2021-11-28 19:34 ` [PATCH 07/12] habanalabs: add power information type to POWER_GET packet Oded Gabbay
2021-11-28 19:34 ` [PATCH 08/12] habanalabs: change misleading IRQ warning during reset Oded Gabbay
2021-11-28 19:34 ` [PATCH 09/12] habanalabs: handle events during soft-reset Oded Gabbay
2021-11-28 19:34 ` Oded Gabbay [this message]
2021-11-28 19:34 ` [PATCH 11/12] habanalabs: add SOB information to signal submission uAPI Oded Gabbay
2021-11-28 19:34 ` [PATCH 12/12] habanalabs: enable access to info ioctl during hard reset Oded Gabbay

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20211128193435.266534-10-ogabbay@kernel.org \
    --to=ogabbay@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=osharabi@habana.ai \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®