From: Ian Rogers <irogers@google.com>
To: Kashyap Desai <kashyap.desai@broadcom.com>,
Sumit Saxena <sumit.saxena@broadcom.com>,
Shivasharan S <shivasharan.srikanteshwara@broadcom.com>,
Chandrakanth patil <chandrakanth.patil@broadcom.com>,
"James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>,
"Martin K. Petersen" <mkp@kernel.org>
Cc: megaraidlinux.pdl@broadcom.com, linux-scsi@vger.kernel.org,
linux-kernel@vger.kernel.org, llvm@lists.linux.dev,
Elkin Cruz <elkin@google.com>, Ian Rogers <irogers@google.com>
Subject: [PATCH] scsi: megaraid: Fix uninitialized DMA handle and passthru length TOCTOU
Date: Thu, 19 Nov 2026 09:05:22 -0800 [thread overview]
Message-ID: <20261119170522.1516239-1-irogers@google.com> (raw)
In megadev_ioctl(), data_dma_hndl is declared uninitialized. On the DCMD
path, when uioc.xferlen == 0, dma_alloc_coherent() is skipped, yet
mc.xferaddr = (u32)data_dma_hndl unconditionally writes 32 bits of
uninitialized kernel stack into the mailbox command sent to the adapter.
On the MEGA_MBOXCMD_PASSTHRU path, pthru resides in coherent DMA memory
shared with the controller, and pthru->dataxferlen is re-read after
mega_internal_command() for both copy_to_user() and dma_free_coherent().
If the device modifies pthru->dataxferlen in coherent memory, the cleanup
path can over-read data or pass a mismatched size or uninitialized
data_dma_hndl to dma_free_coherent().
Initialize data_dma_hndl to 0 and snapshot pthru->dataxferlen into a local
variable once copied from userspace.
Found via an x86_64 LLVM=1 (Clang 21.1.8) allyesconfig audit with
-Wconditional-uninitialized enabled; fix and changelog drafted with an AI
coding assistant. Build-tested under LLVM=1 W=1 and
-Wconditional-uninitialized; not tested on legacy MegaRAID hardware.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Ian Rogers <irogers@google.com>
---
drivers/scsi/megaraid.c | 25 ++++++++++++++-----------
1 file changed, 14 insertions(+), 11 deletions(-)
diff --git a/drivers/scsi/megaraid.c b/drivers/scsi/megaraid.c
index a4079866f747..970e9b2ec781 100644
--- a/drivers/scsi/megaraid.c
+++ b/drivers/scsi/megaraid.c
@@ -2966,7 +2966,7 @@ megadev_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
mega_passthru *pthru; /* copy user passthru here */
dma_addr_t pthru_dma_hndl;
void *data = NULL; /* data to be transferred */
- dma_addr_t data_dma_hndl; /* dma handle for data xfer area */
+ dma_addr_t data_dma_hndl = 0; /* dma handle for data xfer area */
megacmd_t mc;
#if MEGA_HAVE_STATS
megastat_t __user *ustats = NULL;
@@ -3137,6 +3137,7 @@ megadev_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
/* Is it a passthru command or a DCMD */
if( uioc.uioc_rmbox[0] == MEGA_MBOXCMD_PASSTHRU ) {
/* Passthru commands */
+ u32 dataxferlen;
pthru = dma_alloc_coherent(&pdev->dev,
sizeof(mega_passthru),
@@ -3170,9 +3171,10 @@ megadev_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
/*
* Is there a data transfer
*/
- if( pthru->dataxferlen ) {
+ dataxferlen = pthru->dataxferlen;
+ if (dataxferlen) {
data = dma_alloc_coherent(&pdev->dev,
- pthru->dataxferlen,
+ dataxferlen,
&data_dma_hndl,
GFP_KERNEL);
@@ -3199,12 +3201,13 @@ megadev_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
/*
* Is data coming down-stream
*/
- if( pthru->dataxferlen && (uioc.flags & UIOC_WR) ) {
+ if (dataxferlen && (uioc.flags & UIOC_WR)) {
/*
* Get the user data
*/
- if( copy_from_user(data, (char __user *)(unsigned long) uxferaddr,
- pthru->dataxferlen) ) {
+ if (copy_from_user(data,
+ (char __user *)(unsigned long)uxferaddr,
+ dataxferlen)) {
rval = (-EFAULT);
goto freemem_and_return;
}
@@ -3228,9 +3231,9 @@ megadev_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
/*
* Is data going up-stream
*/
- if( pthru->dataxferlen && (uioc.flags & UIOC_RD) ) {
- if( copy_to_user((char __user *)(unsigned long) uxferaddr, data,
- pthru->dataxferlen) ) {
+ if (dataxferlen && (uioc.flags & UIOC_RD)) {
+ if (copy_to_user((char __user *)(unsigned long)uxferaddr,
+ data, dataxferlen)) {
rval = (-EFAULT);
}
}
@@ -3244,9 +3247,9 @@ megadev_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
rval = -EFAULT;
freemem_and_return:
- if( pthru->dataxferlen ) {
+ if (dataxferlen) {
dma_free_coherent(&pdev->dev,
- pthru->dataxferlen, data,
+ dataxferlen, data,
data_dma_hndl);
}
--
2.56.1.708.g77570a6a1d-goog
reply other threads:[~2026-11-19 17:05 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
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=20261119170522.1516239-1-irogers@google.com \
--to=irogers@google.com \
--cc=James.Bottomley@HansenPartnership.com \
--cc=chandrakanth.patil@broadcom.com \
--cc=elkin@google.com \
--cc=kashyap.desai@broadcom.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=llvm@lists.linux.dev \
--cc=megaraidlinux.pdl@broadcom.com \
--cc=mkp@kernel.org \
--cc=shivasharan.srikanteshwara@broadcom.com \
--cc=sumit.saxena@broadcom.com \
/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®