CVE-2026-64339

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> usb: misc: usbio: bound bulk IN response length to the received transfer<br /> <br /> usbio_bulk_msg() copies bpkt_len = le16_to_cpu(bpkt-&gt;len) bytes out of<br /> the bulk IN buffer (usbio-&gt;rxbuf, allocated with size usbio-&gt;rxbuf_len)<br /> into the caller&amp;#39;s buffer. bpkt_len is fully controlled by the device<br /> and is only checked against ibuf_len; ibuf_len in turn is checked<br /> against usbio-&gt;txbuf_len, not against rxbuf_len:<br /> <br /> if ((obuf_len &gt; (usbio-&gt;txbuf_len - sizeof(*bpkt))) ||<br /> (ibuf_len &gt; (usbio-&gt;txbuf_len - sizeof(*bpkt))))<br /> return -EMSGSIZE;<br /> <br /> txbuf_len and rxbuf_len are taken independently from the bulk OUT and<br /> bulk IN endpoint wMaxPacketSize in usbio_probe(). A malicious or<br /> malfunctioning device that advertises a large bulk OUT endpoint and a<br /> small bulk IN endpoint (e.g. by claiming one of the quirk-free IDs such<br /> as the Lattice NX33U, 0x2ac1:0x20cb) therefore makes ibuf_len, and<br /> hence the device-supplied bpkt_len, exceed rxbuf_len. memcpy() then<br /> reads up to txbuf_len - rxbuf_len bytes past the end of the rxbuf slab<br /> object. The over-read bytes are handed back to the i2c layer and on to<br /> user space through i2c-dev, disclosing adjacent slab memory; with KASAN<br /> this is reported as a slab-out-of-bounds read.<br /> <br /> The number of bytes actually received is already known: act equals the<br /> URB actual_length and is bounded by rxbuf_len. Reject any response<br /> that claims more payload than was received, mirroring the existing<br /> "act