linux-kernel - Re: [PATCH] char: tpm: tpm-buf: Fix uninitialized return values in read helpers

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: <Z_eIAOSZZv4zJWVO@kernel.org>
Date: Thu, 10 Apr 2025 11:57:36 +0300
From: Jarkko Sakkinen <jarkko@...nel.org>
To: Purva Yeshi <purvayeshi550@...il.com>
Cc: Stefano Garzarella <sgarzare@...hat.com>, peterhuewe@....de,
	jgg@...pe.ca, linux-integrity@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: Re: [PATCH] char: tpm: tpm-buf: Fix uninitialized return values in
 read helpers

On Thu, Apr 10, 2025 at 11:55:22AM +0300, Jarkko Sakkinen wrote:
> On Thu, Apr 10, 2025 at 02:12:07PM +0530, Purva Yeshi wrote:
> > On 10/04/25 13:21, Stefano Garzarella wrote:
> > > On Thu, Apr 10, 2025 at 09:14:58AM +0300, Jarkko Sakkinen wrote:
> > > > On Thu, Apr 10, 2025 at 02:25:36AM +0530, Purva Yeshi wrote:
> > > > > Fix Smatch-detected error:
> > > > > drivers/char/tpm/tpm-buf.c:208 tpm_buf_read_u8() error:
> > > > > uninitialized symbol 'value'.
> > > > > drivers/char/tpm/tpm-buf.c:225 tpm_buf_read_u16() error:
> > > > > uninitialized symbol 'value'.
> > > > > drivers/char/tpm/tpm-buf.c:242 tpm_buf_read_u32() error:
> > > > > uninitialized symbol 'value'.
> > > > > 
> > > > > Call tpm_buf_read() to populate value but do not check its return
> > > > > status. If the read fails, value remains uninitialized, causing
> > > > > undefined behavior when returned or processed.
> > > > > 
> > > > > Initialize value to zero to ensure a defined return even if
> > > > > tpm_buf_read() fails, avoiding undefined behavior from using
> > > > > an uninitialized variable.
> > > > 
> > > > How does tpm_buf_read() fail?
> > > 
> > > If TPM_BUF_BOUNDARY_ERROR is set (or we are setting it), we are
> > > effectively returning random stack bytes to the caller.
> > > Could this be a problem?
> > > 
> > > If it is, maybe instead of this patch, we could set `*output` to zero in
> > > the error path of tpm_buf_read(). Or return an error from tpm_buf_read()
> > > so callers can return 0 or whatever they want.
> > > 
> > > Thanks,
> > > Stefano
> > > 
> > 
> > Hi Jarkko, Stefano,
> > Thank you for the review.
> > 
> > I've revisited the issue and updated the implementation of tpm_buf_read() to
> > zero out the *output buffer in the error paths, instead of initializing the
> > return value in each caller.
> > 
> > static void tpm_buf_read(struct tpm_buf *buf, off_t *offset, size_t count,
> > void *output)
> > {
> > 	off_t next_offset;
> > 
> > 	/* Return silently if overflow has already happened. */
> > 	if (buf->flags & TPM_BUF_BOUNDARY_ERROR) {
> > 		memset(output, 0, count);
> > 		return;
> > 	}
> > 
> > 	next_offset = *offset + count;
> > 	if (next_offset > buf->length) {
> > 		WARN(1, "tpm_buf: read out of boundary\n");
> > 		buf->flags |= TPM_BUF_BOUNDARY_ERROR;
> > 		memset(output, 0, count);
> > 		return;
> > 	}
> > 
> > 	memcpy(output, &buf->data[*offset], count);
> > 	*offset = next_offset;
> > }
> 
> Please don't touch this.

If you want to do anything, check the call sites for raw tpm_buf_read()
instead, which is not very common.

BR, Jarkko