In the Linux kernel, the following vulnerability has been resolved:
ionic: use devconsumeskb_any outside of napi
If we're not in a NAPI softirq context, we need to be careful about how we call napiconsumeskb(), specifically we need to call it with budget==0 to signal to it that we're not in a safe context.
This was found while running some configuration stress testing of traffic and a change queue config loop running, and this curious note popped out:
[ 4371.402645] BUG: using smpprocessorid() in preemptible [00000000] code: ethtool/20545 [ 4371.402897] caller is napiskbcacheput+0x16/0x80 [ 4371.403120] CPU: 25 PID: 20545 Comm: ethtool Kdump: loaded Tainted: G OE 6.10.0-rc3-netnext+ #8 [ 4371.403302] Hardware name: HPE ProLiant DL360 Gen10/ProLiant DL360 Gen10, BIOS U32 01/23/2021 [ 4371.403460] Call Trace: [ 4371.403613] <TASK> [ 4371.403758] dumpstacklvl+0x4f/0x70 [ 4371.403904] checkpreemptiondisabled+0xc1/0xe0 [ 4371.404051] napiskbcacheput+0x16/0x80 [ 4371.404199] ionictxclean+0x18a/0x240 [ionic] [ 4371.404354] ionictxcqservice+0xc4/0x200 [ionic] [ 4371.404505] ionictxflush+0x15/0x70 [ionic] [ 4371.404653] ? ioniclifqcqdeinit.isra.23+0x5b/0x70 [ionic] [ 4371.404805] ionictxrxdeinit+0x71/0x190 [ionic] [ 4371.404956] ionicreconfigurequeues+0x5f5/0xff0 [ionic] [ 4371.405111] ionicsetringparam+0x2e8/0x3e0 [ionic] [ 4371.405265] ethnlsetrings+0x1f1/0x300 [ 4371.405418] ethnldefaultsetdoit+0xbb/0x160 [ 4371.405571] genlfamilyrcvmsg_doit+0xff/0x130 [...]
I found that ionictxclean() calls napiconsumeskb() which calls napiskbcacheput(), but before that last call is the note /* Zero budget indicate non-NAPI context called us, like netpoll */ and DEBUGNETWARNONONCE(!insoftirq());
Those are pretty big hints that we're doing it wrong. We can pass a context hint down through the calls to let ionictxclean() know what we're doing so it can call napiconsumeskb() correctly.