Re: [PATCH] tools/bootconfig: Fix integer overflow in size check

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Incomplete Fix / Integer Truncation Leading to Heap Overflow
--

commit 2d027c4ee8625f17122dea0afb65c1e4b21ae531
Author: Masami Hiramatsu (Google) <mhiramat@kernel.org>

    tools/bootconfig: Fix integer overflow in size check
    
    This commit modifies the size check in load_xbc_from_initrd() to prevent
    an integer overflow on 32-bit systems. It changes the condition to
    subtract BOOTCONFIG_FOOTER_SIZE from stat.st_size instead of adding it
    to the bootconfig size.

    Closes: https://lore.kernel.org/all/20260909161113.16C691F00A3A@smtp.kernel.org/

> diff --git a/tools/bootconfig/main.c b/tools/bootconfig/main.c
> index 7dc9fff9b6370..b380ad6777faa 100644
> --- a/tools/bootconfig/main.c
> +++ b/tools/bootconfig/main.c
> @@ -218,7 +218,7 @@ static int load_xbc_from_initrd(int fd, char **buf)
>  	csum = le32toh(csum);
>  
>  	/* Wrong size error  */
> -	if (stat.st_size < size + BOOTCONFIG_FOOTER_SIZE) {
> +	if (size > stat.st_size - BOOTCONFIG_FOOTER_SIZE) {

[Severity: High]
This is a pre-existing issue, but the fix doesn't seem to completely resolve
the integer truncation and heap overflow vulnerabilities.

Can this size check still be bypassed for large files? Since stat.st_size is a
64-bit off_t, if the file size is larger than 4.29 GB, stat.st_size -
BOOTCONFIG_FOOTER_SIZE can exceed UINT32_MAX. A corrupted uint32_t size of
0xFFFFFFFF would bypass this check because it is always smaller than the
right side.

If bypassed, does passing 0xFFFFFFFF to load_xbc_fd() cause an integer
truncation and heap overflow?

tools/bootconfig/main.c:load_xbc_fd() {
    ...
    static int load_xbc_fd(int fd, char **buf, int size)
    ...
        *buf = malloc(size + 1);
        if (!*buf)
            return -ENOMEM;
    
        ret = read(fd, *buf, size);
    ...
        (*buf)[size] = '\0';
    ...
}

Because size is an int, 0xFFFFFFFF is truncated to -1. This results in
malloc(0) which returns a valid heap chunk. The read() call then receives -1
which gets cast to SIZE_MAX, overflowing the heap chunk. Finally, a null byte
is written to a negative array index (-1), causing an out-of-bounds write.

Does this same vulnerability also affect load_xbc_file() when processing
standalone files >= 2GB?

tools/bootconfig/main.c:load_xbc_file() {
    ...
    ret = fstat(fd, &stat);
    ...
    ret = load_xbc_fd(fd, buf, stat.st_size);
    ...
}

This implicitly truncates the 64-bit off_t stat.st_size into the 32-bit int
size parameter of load_xbc_fd().

>  		pr_err("bootconfig size is too big\n");
>  		return -E2BIG;
>  	}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/178900024679.183840.10345859264677533824.stgit@devnote2?part=1