Wind River Support Network

Description

In the Linux kernel, the following vulnerability has been resolved:net: dsa: improve shutdown sequenceAlexander Sverdlin presents 2 problems during shutdown with thelan9303 driver. On
e is specific to lan9303 and the other just happensto reproduce there.The first problem is that lan9303 is unique among DSA drivers in that itcalls dev_get_drvdata() at "arbitrary run
time" (not probe, not shutdown,not remove):phy_state_machine()-> ...   -> dsa_user_phy_read()      -> ds->ops->phy_read()         -> lan9303_phy_read()            -> chip->ops->phy_re
ad()               -> lan9303_mdio_phy_read()                  -> dev_get_drvdata()But we never stop the phy_state_machine(), so it may continue to runafter dsa_switch_shutdown(). Our
 common pattern in all DSA drivers isto set drvdata to NULL to suppress the remove() method that may comeafterwards. But in this case it will result in an NPD.The second problem is th
at the way in which we setdp->conduit->dsa_ptr = NULL; is concurrent with receive packetprocessing. dsa_switch_rcv() checks once whether dev->dsa_ptr is NULL,but afterwards, rather th
an continuing to use that non-NULL value,dev->dsa_ptr is dereferenced again and again without NULL checks:dsa_conduit_find_user() and many other places. In between dereferences,there 
is no locking to ensure that what was valid once continues to bevalid.Both problems have the common aspect that closing the conduit interfacesolves them.In the first case, dev_close(c
onduit) triggers the NETDEV_GOING_DOWNevent in dsa_user_netdevice_event() which closes user ports as well.dsa_port_disable_rt() calls phylink_stop(), which synchronously stopsthe phyl
ink state machine, and ds->ops->phy_read() will thus no longercall into the driver after this point.In the second case, dev_close(conduit) should do this, as perDocumentation/networki
ng/driver.rst:| Quiescence| ----------|| After the ndo_stop routine has been called, the hardware must| not receive or transmit any data.  All in flight packets must| be aborted. If n
ecessary, poll or wait for completion of| any reset commands.So it should be sufficient to ensure that later, when we zeroizeconduit->dsa_ptr, there will be no concurrent dsa_switch_r
cv() callon this conduit.The addition of the netif_device_detach() function is to ensure thatioctls, rtnetlinks and ethtool requests on the user ports no longerpropagate down to the d
river - we're no longer prepared to handle them.The race condition actually did not exist when commit 0650bf52b31f("net: dsa: be compatible with masters which unregister on shutdown")
first introduced dsa_switch_shutdown(). It was created later, when westopped unregistering the user interfaces from a bad spot, and we justreplaced that sequence with a racy zeroizati
on of conduit->dsa_ptr(one which doesn't ensure that the interfaces aren't up).

CVEs

• CVE-2024-49998