From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-alma10-1.taild15c8.ts.net [100.103.45.18]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 8FAAA4756AE; Tue, 28 Jul 2026 21:56:52 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=100.103.45.18 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1785275820; cv=none; b=IiwN+rfkZ5NMR+mZAazodL6iJrWiKragVHMO7VegfHB5oTXDqjHUywUtbHuTP+xbx4j8d45Hlss5MV/8m6voJZeS00oq9MVEScYaIdWOTvROkSR25AaXHqVQV9KPre9n7lni8bFMBygUBLcM6ElsI7WgffVwnAEAE17NyX+1LiQ= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1785275820; c=relaxed/simple; bh=cdy0g/nKdxty366qImqvpcAmcZhy+zBh0Z2t6QEgwRg=; h=Date:From:To:Cc:Subject:Message-ID:MIME-Version:Content-Type: Content-Disposition:In-Reply-To; b=ixr1N8VQNKTvUbVUBum/JjskjAbIkjmc9k39xYGsKx+SXLd1+Syjk5ipSroXyi2ovG1smR8eybtmF05YF/haueUupoVnuEzyEF00jVdD3koS9w8NqLPsGcNhQD/oEdS0bBPYJ2jk1rsppt0ZyT8ed8KC9VfHw70A2bNDIxsvUyA= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=Qt2HWEaC; arc=none smtp.client-ip=100.103.45.18 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="Qt2HWEaC" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 4DF661F000E9; Tue, 28 Jul 2026 21:56:48 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=kernel.org; s=k20260515; t=1785275808; bh=D/kVLOj+hFIJeBH6n2ip5Hjahw38qkedPguWVx0z17o=; h=Date:From:To:Cc:Subject:In-Reply-To; b=Qt2HWEaCu0Jw/iwYyCkRcmuRQvZHh94V9HvLCQ+lT53xGJypy+bQGSot985RZ5JjD 8zFG3nGVHf5je3qu4zjqtgWkdx7PRpLCNY+Wsk435TpmfU/HCY1yY1Aw30Nft7yg/C Mu4pJngQbpsVXzRp5HOpXtmReUlKOMeXl1Cb4KO4IsSF0NbJj7jz3ftH+p1JrCoJXP JAy7d8VSbckvXqRqBFQclEMZc0YJWJ/WeH2SI+Og9m0fw9MByGIFNtMsyntkMNLSpt WdXao1biO4+cNfFx6HVVqT5JwwhJotIJ3Wo5Wb8HyJvO3Mhizpxvwb+If9yxLBMkI9 TtCQuhWZVZo1A== Date: Tue, 28 Jul 2026 16:56:47 -0500 From: Bjorn Helgaas To: Deepanshu Kartikey Cc: bhelgaas@google.com, kw@linux.com, linux-pci@vger.kernel.org, linux-kernel@vger.kernel.org, syzbot+c7604c9fdd7580cca4e0@syzkaller.appspotmail.com Subject: Re: [PATCH v2] PCI/proc: check user access return values in proc_bus_pci_{read,write}() Message-ID: <20260728215647.GA1402909@bhelgaas> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20260504022231.15501-1-kartikey406@gmail.com> On Mon, May 04, 2026 at 07:52:31AM +0530, Deepanshu Kartikey wrote: > proc_bus_pci_write() ignores the return value of __get_user(). On a > faulting user pointer the extable fixup zeros the destination, and the > function writes those zeros to PCI configuration space. > > syzbot triggers this with writev()-ing a NULL iov_base to > /proc/bus/pci/00/03.0 (the virtio-blk controller in the syzkaller VM): > zero is written to the Command register, clearing Bus Master Enable, > and the disk stops responding. In-flight journal writes never complete > and jbd2 hangs in wait_on_buffer() indefinitely: > > INFO: task jbd2/sda1-8 blocked in I/O wait for more than 143 seconds. > __wait_on_buffer fs/buffer.c:123 > jbd2_journal_commit_transaction+0x388a/0x6870 fs/jbd2/commit.c:837 > kjournald2 fs/jbd2/journal.c:201 > > proc_bus_pci_read() has the symmetric problem with __put_user(): a > faulting user pointer silently drops config-space data and returns > success. > > Switch both functions to get_user()/put_user(), which combine the > access_ok() check with the load/store and return -EFAULT on failure. > The up-front access_ok() can be removed accordingly. On error, jump to > a common label that releases the runtime-PM reference and returns > -EFAULT. > > Reported-by: syzbot+c7604c9fdd7580cca4e0@syzkaller.appspotmail.com > Closes: https://syzkaller.appspot.com/bug?extid=c7604c9fdd7580cca4e0 > Tested-by: syzbot+c7604c9fdd7580cca4e0@syzkaller.appspotmail.com > Signed-off-by: Deepanshu Kartikey Thanks for this patch! I asked gemini to reduce the syzbot repro details and focus on the issues of writing bogus data to config space, returning success with invalid read data, and returning -EFAULT. It suggested the log below; what do you think? PCI/proc: Return -EFAULT on get_user() and put_user() failures proc_bus_pci_write() ignores the return value of __get_user(). On a faulting user-space pointer, the exception table fixup zeros the destination, and proc_bus_pci_write() writes those invalid zeros into PCI configuration space. proc_bus_pci_read() ignores the return value of __put_user(). On a faulting user-space pointer, it fails to copy the PCI configuration space data to user space but still returns success. Fix both issues by: - Switching to get_user() and put_user() and checking return values - Returning -EFAULT, not -EINVAL, when get_user() or put_user() fails - Removing the redundant access_ok() checks, since get_user() and put_user() validate access internally > --- > Changes in v2: > - Use get_user()/put_user() and drop access_ok() (Krzysztof) > - Rename label to err: per kernel convention (Krzysztof) > - Simplify error path to release runtime-PM and return -EFAULT (Krzysztof) > - Apply the same fix to proc_bus_pci_read() (Krzysztof) > --- > drivers/pci/proc.c | 44 ++++++++++++++++++++++++++++---------------- > 1 file changed, 28 insertions(+), 16 deletions(-) > > diff --git a/drivers/pci/proc.c b/drivers/pci/proc.c > index ce36e35681e8..8e624d829840 100644 > --- a/drivers/pci/proc.c > +++ b/drivers/pci/proc.c > @@ -53,15 +53,13 @@ static ssize_t proc_bus_pci_read(struct file *file, char __user *buf, > nbytes = size - pos; > cnt = nbytes; > > - if (!access_ok(buf, cnt)) > - return -EINVAL; > - > pci_config_pm_runtime_get(dev); > > if ((pos & 1) && cnt) { > unsigned char val; > pci_user_read_config_byte(dev, pos, &val); > - __put_user(val, buf); > + if (put_user(val, buf)) > + goto err; > buf++; > pos++; > cnt--; > @@ -70,7 +68,8 @@ static ssize_t proc_bus_pci_read(struct file *file, char __user *buf, > if ((pos & 3) && cnt > 2) { > unsigned short val; > pci_user_read_config_word(dev, pos, &val); > - __put_user(cpu_to_le16(val), (__le16 __user *) buf); > + if (put_user(cpu_to_le16(val), (__le16 __user *) buf)) > + goto err; > buf += 2; > pos += 2; > cnt -= 2; > @@ -79,7 +78,8 @@ static ssize_t proc_bus_pci_read(struct file *file, char __user *buf, > while (cnt >= 4) { > unsigned int val; > pci_user_read_config_dword(dev, pos, &val); > - __put_user(cpu_to_le32(val), (__le32 __user *) buf); > + if (put_user(cpu_to_le32(val), (__le32 __user *) buf)) > + goto err; > buf += 4; > pos += 4; > cnt -= 4; > @@ -89,7 +89,8 @@ static ssize_t proc_bus_pci_read(struct file *file, char __user *buf, > if (cnt >= 2) { > unsigned short val; > pci_user_read_config_word(dev, pos, &val); > - __put_user(cpu_to_le16(val), (__le16 __user *) buf); > + if (put_user(cpu_to_le16(val), (__le16 __user *) buf)) > + goto err; > buf += 2; > pos += 2; > cnt -= 2; > @@ -98,7 +99,8 @@ static ssize_t proc_bus_pci_read(struct file *file, char __user *buf, > if (cnt) { > unsigned char val; > pci_user_read_config_byte(dev, pos, &val); > - __put_user(val, buf); > + if (put_user(val, buf)) > + goto err; > pos++; > } > > @@ -106,6 +108,10 @@ static ssize_t proc_bus_pci_read(struct file *file, char __user *buf, > > *ppos = pos; > return nbytes; > + > +err: > + pci_config_pm_runtime_put(dev); > + return -EFAULT; > } > > static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > @@ -129,14 +135,12 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > nbytes = size - pos; > cnt = nbytes; > > - if (!access_ok(buf, cnt)) > - return -EINVAL; > - > pci_config_pm_runtime_get(dev); > > if ((pos & 1) && cnt) { > unsigned char val; > - __get_user(val, buf); > + if (get_user(val, buf)) > + goto err; > pci_user_write_config_byte(dev, pos, val); > buf++; > pos++; > @@ -145,7 +149,8 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > > if ((pos & 3) && cnt > 2) { > __le16 val; > - __get_user(val, (__le16 __user *) buf); > + if (get_user(val, (__le16 __user *) buf)) > + goto err; > pci_user_write_config_word(dev, pos, le16_to_cpu(val)); > buf += 2; > pos += 2; > @@ -154,7 +159,8 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > > while (cnt >= 4) { > __le32 val; > - __get_user(val, (__le32 __user *) buf); > + if (get_user(val, (__le32 __user *) buf)) > + goto err; > pci_user_write_config_dword(dev, pos, le32_to_cpu(val)); > buf += 4; > pos += 4; > @@ -163,7 +169,8 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > > if (cnt >= 2) { > __le16 val; > - __get_user(val, (__le16 __user *) buf); > + if (get_user(val, (__le16 __user *) buf)) > + goto err; > pci_user_write_config_word(dev, pos, le16_to_cpu(val)); > buf += 2; > pos += 2; > @@ -172,7 +179,8 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > > if (cnt) { > unsigned char val; > - __get_user(val, buf); > + if (get_user(val, buf)) > + goto err; > pci_user_write_config_byte(dev, pos, val); > pos++; > } > @@ -182,6 +190,10 @@ static ssize_t proc_bus_pci_write(struct file *file, const char __user *buf, > *ppos = pos; > i_size_write(ino, dev->cfg_size); > return nbytes; > + > +err: > + pci_config_pm_runtime_put(dev); > + return -EFAULT; > } > > #ifdef HAVE_PCI_MMAP > -- > 2.43.0 >