Is the co-pilot itself performance-sensitive? [LWN.net]

Is the co-pilot itself performance-sensitive?

Posted Sep 23, 2024 2:03 UTC (Mon) by viro (subscriber, #7872)
In reply to: Is the co-pilot itself performance-sensitive? by ms-tg
Parent article: pcp: pmcd network daemon review (SUSE Security Team Blog)

>> if (vindex < 0 || vindex > pdulen / sizeof(__pmPDU)) {

> Question: is it worth someone making a tiny PR with this change and the explanation above?

Wouldn't vindex equal to pdulen / sizeof(....) end up with off-by-one?

to post comments

Posted Sep 23, 2024 9:12 UTC (Mon) by rweikusat2 (subscriber, #117920) [Link] (1 responses)

It would. The best way to express this without errors is probably

if (vindex < 0 || vindex >= pdulen / sizeof(*pdubuf)) {

as this uses the actual type of the array element, whatever it is.

Posted Sep 23, 2024 9:25 UTC (Mon) by rweikusat2 (subscriber, #117920) [Link]

Additional remark: Assuming that pdubuf if either an array of __pmPDU or a pointer to __pmPDU, the expressions

pdubuf[0]

and

*pdubuf

are semantically identical.