Merge tag 'iwlwifi-for-kalle-2018-08-09' of git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/iwlwifi-fixes

Some more fixes for 4.13

* Fix a memory leak in the SAR code;
* Fix a stuck queue case in AP mode;
* Convert a WARN to a simple debug in a legitimate race case (from
  which we can recover);
* Fix a severe throughput aggregation on 9000-family devices due to
  aggregation issues.

When an RX block-ack session times out, the firmware, which offloads
RX reordering but not the BA session negotiation, stops the session
but doesn't send a DELBA.  This causes the the session to remain
active in the remote device, so no more BA sessions will be
established, causing a severe throughput degradation due to the lack
of aggregation.

Use the new ieee80211_rx_ba_timer_expired API when the ba session timer
expires, since this will tear down the ba session and also send a delba.

The previous API used is intended for drivers that offload the
addba/delba negotiation, but not the rx reordering, while our driver
does the opposite.

This patch depends on "mac80211: add api to start ba session timer
expired flow".

Signed-off-by: Naftali Goldstein <naftali.goldstein@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Some drivers handle rx buffer reordering internally (and by extension
handle also the rx ba session timer internally), but do not ofload the
addba/delba negotiation.
Add an api for these drivers to properly tear-down the ba session,
including sending a delba.

Signed-off-by: Naftali Goldstein <naftali.goldstein@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

When we start an Rx A-MPDU session, we first get the AddBA
request, then we send an ADD_STA command to the firmware
that will reply with a BAID which is a hardware resource
that tracks the BA session.
This BAID will appear on each and every frame that we get
from the firwmare until the A-MPDU session is torn down.
In the Rx path, we look at this BAID to manage the
reordering buffer.

This flow is inherently racy since the hardware will start
to put the BAID in the frames it receives even if the
firmware hasn't sent the response to the ADD_STA command.
This basically means that the driver can get frames with
a valid BAID that it doesn't know yet.
When that happens, the driver used to WARN.
Fix this by simply not WARN in this case. When the driver
will know abou the BAID, it will initialise the relevant
states and the next frame with a valid BAID will refresh
them.

Fixes: b915c101

 ("iwlwifi: mvm: add reorder buffer per queue")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

In AP mode, if a station is removed just as it is adding a new stream,
the queue in question will remain stopped and no more TX will happen
in this queue, leading to connection failures and other problems.

This is because under DQA, when tx is deferred because a queue needs
to be allocated, the mac queue for that TID is stopped until the new
stream is added.  If at this point the station that this stream
belongs to is removed, all the deferred tx frames are purged, but the
mac queue is not restarted. As a result, all following tx on this
queue will not be transmitted.

Fix this by starting the relevant mac queues when the deferred tx
frames are purged.

Fixes: 24afba76

 ("iwlwifi: mvm: support bss dynamic alloc/dealloc of queues")
Signed-off-by: Avraham Stern <avraham.stern@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

We should free 'wgds.pointer' here as done a few lines above in another
error handling path.
It was allocated within 'acpi_evaluate_object()'.

Fixes: c52030a01ccc ("iwlwifi: mvm: add GEO_TX_POWER_LIMIT cmd for geographic tx power table")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Merge tag 'iwlwifi-for-kalle-2017-08-02' of git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/iwlwifi-fixes

Some fixes in iwlwifi for 4.13

* Some simple PCI HW ID fix-ups and additions for family 9000;
* A couple of bugzilla fixes:
  - Remove a bogus warning message with new FWs (196915)
  - Don't allow illegal channel options to be used (195299)
* A fix for checksum offload in family 9000;
* A fix serious throughput degradation in 11ac with multiple streams;
* An old bug in SMPS where the firmware was not aware of SMPS changes;

Add two PCI IDs for the 9160 series.
Add five PCI IDs for the 9260 series.
Add one PCI IDs for the 9270 series.
Add seven PCI IDs for the 9460 series.
Add five PCI IDs for the 9560 series.

Signed-off-by: Tzipi Peres <tzipi.peres@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Set the STA_FLG_RTS_MIMO_PROT bit in station_flags_msk of the add sta
command, so that when smps mode changes, the FW will know about it.

In particular, in AP mode, clients are added upon receival of an auth
request, at which point there's no knowledge of the client's smps mode.
When the assoc request arrives, the add_sta command is resent to modify
the station parameters. At this point the driver knows the smps mode,
but since the corresponding bit in the mask is not set, the fw doesn't
update this field so there's no rts protection for mimo.

Fixes: 5bc5aaad

 ("iwlwifi: mvm: set up initial SMPS/NSS station info")
Signed-off-by: Naftali Goldstein <naftali.goldstein@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

C step NICs should use the latest FW (currently B step).
Correct the condition to make C step NICs advanced its default FW name
to the latest one.
Also rename _next_ to b_or_c to avoid confusion.

Fixes: 5da083d1

 ("iwlwifi: add support for 9000 HW B-step NICs")
Signed-off-by: Haim Dreyfuss <haim.dreyfuss@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Statistics should be collected according to the actual rate a
frame/aggregation was transmitted and not according to the initial rate
from the last LQ command (these rates are different if the frames were
retransmitted at a lower rate from the rate scale table).

This is needed to remove throughput degradation.

Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

The bit was set only if there was at least one reclaimed frame in an
aggregation. It's important to set it also in the case that the whole
A-MPDU was lost, otherwise rate scaling statistics will not be
updated correctly. Thus, set it always in ba notification handler.

This fixes a throughput degradation of about 20% in certain scenarios
with multiple streams on 11ac.

Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

When we create a regulatory domain out of an MCC
notification, we need to make sure that all the channels
in the rule have the exact same properties.
The current code mixes channel 36 and 40 although 36 can be
a control channel with HT40+ (36, 40) whereas 40 can't be
a control channel with HT40+ since  (40, 44) is invalid.

Because of that, cfg80211 would allow to connect in 40MHz
to APs that are configured to channel 40 HT40+ and that made
our firmware assert.

Fix this by checking the bandwidth flags before taking the
decision if the rule should be split.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=195299 partly.

Fixes: af45a900

 ("iwlwifi: create regdomain from mcc_update_cmd response")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

The firmware now adds a new DWORD for the MLME offload's
capability even on firmware versions that don't support
it.
Add the TLV bit to avoid getting the print:
capa flags index 3 larger than supported by driver.

This fixes the bugzilla:
https://bugzilla.kernel.org/show_bug.cgi?id=196195



Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

When we enabled TCP checksum offload, we need to tell the
firmware where the IP header starts. If we have an IV, then
we need to adapt that value since the IV is placed before
the SNAP header. This is true only for cases where the
driver adds the IV, not the WEP case in which the IV is
added by the firmware itself.

On A000 devices series, the IV is always added by the
device.

Fix this.

Fixes: 5e6a98dc

 ("iwlwifi: mvm: enable TCP/UDP checksum support for 9000 family")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Due to a bugfix in wireless tree and the commit mentioned below a merge
was needed which went haywire. So the submitted change resulted in the
function brcmf_sdiod_sgtable_alloc() being called twice during the probe
thus leaking the memory of the first call.

Cc: stable@vger.kernel.org # 4.6.x
Fixes: 4d792895

 ("brcmfmac: switch to new platform data")
Reported-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Reviewed-by: Hante Meuleman <hante.meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

The commit to rework the headroom check in start_xmit() now calls
pxskb_expand_head() unconditionally if the header is CoW. Unfortunately,
it does so with the delta between the extant headroom and the header
length, which may be negative if there is already sufficient headroom.

pskb_expand_head() does allow for size being 0, in which case it just
copies, so clamp the header delta to zero.

Opening Chrome (and all my tabs) on a PCIE device was enough to reliably
hit this.

Fixes: 270a6c1f

 ("brcmfmac: rework headroom check in .start_xmit()")
Signed-off-by: Daniel Stone <daniels@collabora.com>
Cc: Arend Van Spriel <arend.vanspriel@broadcom.com>
Cc: James Hughes <james.hughes@raspberrypi.org>
Cc: Hante Meuleman <hante.meuleman@broadcom.com>
Cc: Pieter-Paul Giesberts <pieter-paul.giesberts@broadcom.com>
Cc: Franky Lin <franky.lin@broadcom.com>
Tested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Merge tag 'iwlwifi-for-kalle-2017-07-21' of git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/iwlwifi-fixes

Some iwlwifi fixes for 4.13:

* A few NULL pointer dereferences in the recovery flow;
* A small but important fix for IBSS;
* A one-liner fix for tracing, which was including too much data;
* Some of these are bugzilla bug fixes;

A hardware/firmware error may happen at any point in time. In
particular, it might happen while mac80211 is in the middle of
a flow. We observed the following situation:
 * mac80211 is in authentication flow, in ieee80211_prep_connection()
 * iwlwifi firmware crashes, but no error can be reported at this
   precise point (mostly because the driver method is void, but even
   if it wasn't we'd just shift to a race condition)
 * mac80211 continues the flow, trying to add the AP station
 * iwlwifi has already set its internal restart flag, and so thinks
   that adding the station is part of the restart and already set up,
   so it uses the information that's supposed to already be in the
   struct

This can happen with any flow in mac80211 and with any information
we try to preserve across hardware restarts.

To fix this, only set a new HW_RESTART_REQUESTED flag and translate
that to IN_HW_RESTART once mac80211 actually starts the restart by
calling our start() method. As a consequence, any mac80211 flow in
progress at the time of the restart will properly finish (certainly
with errors), before the restart is attempted.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=195299

.

Reported-by: djagoo <dev@djagoo.io>
Reported-by: Łukasz Siudut <lsiudut@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

When IBSS was implemented for DQA, we missid a few places where it
should be handled in the same way as AP.

Fixes: ee48b722

 ("iwlwifi: mvm: support ibss in dqa mode")
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

iwl_trace_data is somewhat confusing. It returns a bool
that tells if the payload of the skb should be added to
the tx_data event. If it returns false, then the payload
of the skb is added to the tx event.

The purpose is to be able to start tracing with
-e iwlwifi
and record non-data packets only which saves bandwidth.

Since EAPOLs are important, seldom and not real data
packet (despite being WiFi data packets), they are
included in tx event and thus iwl_trace_data returns false
on those. This last part was buggy, and because of that,
all the data packets were included in the tx event.

Fix that.

Fixes: 0c4cb731

 ("iwlwifi: tracing: decouple from mac80211")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

We don't set the error code here so we end up returning ERR_PTR(0) which
is NULL.  The caller doesn't expect that so it results in a NULL
dereference.

Fixes: 2e5d4a8f

 ("iwlwifi: pcie: Add new configuration to enable MSIX")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Sometimes, we can have an firmware crash while trying to
recover from a previous firmware problem.
When that happens, lots of things can go wrong. For example
the stations don't get added properly to mvm->fw_id_to_mac_id.

Mac80211 tries to stop A-MPDU upon reconfig but in case of
a firmware crash we will bail out fairly early and in the
end, we won't delete the A-MPDU Rx timeout.
When that timer expired after a double firmware crash,
we end up dereferencing mvm->fw_id_to_mac_id[sta_id]
which is NULL.

Fixes: 10b2b201

 ("iwlwifi: mvm: add infrastructure for tracking BA session in driver")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

Before TVQM, all TX queues were allocated straight at init.
With TVQM, queues are allocated on demand and hence we need
to check if a queue exists before dereferencing it.

Fixes: 66128fa08806 ("iwlwifi: move to TVQM mode")
Signed-off-by: Mordechai Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

iwlagn_check_ratid_empty takes the tid as a parameter, but
it doesn't check that it is not IWL_TID_NON_QOS.
Since IWL_TID_NON_QOS = 8 and iwl_priv::tid_data is an array
with 8 entries, accessing iwl_priv::tid_data[IWL_TID_NON_QOS]
is a bad idea.
This happened in iwlagn_rx_reply_tx. Since
iwlagn_check_ratid_empty is relevant only to check whether
we can open A-MPDU, this flow is irrelevant if tid is
IWL_TID_NON_QOS. Call iwlagn_check_ratid_empty only inside
the
	if (tid != IWL_TID_NON_QOS)

a few lines earlier in the function.

Cc: <stable@vger.kernel.org>
Reported-by: Seraphime Kirkovski <kirkseraph@gmail.com>
Tested-by: Seraphime Kirkovski <kirkseraph@gmail.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Luca Coelho <luciano.coelho@intel.com>

This reverts commit f95d95a7.

With commit f95d95a7 ("rtlwifi: btcoex: rtl8723be: fix ant_sel not
work"), the kernel has a NULL pointer dereference oops. This content and
the proper fix will be included in a later patch.

Fixes: f95d95a7

 ("rtlwifi: btcoex: rtl8723be: fix ant_sel not work")
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Ping-Ke Shih <pkshih@realtek.com>
Cc: Yan-Hsuan Chuang <yhchuang@realtek.com>
Cc: Birming Chiu <birming@realtek.com>
Cc: Shaofu <shaofu@realtek.com>
Cc: Steven Ting <steventing@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Recent change in brcmf_sdio_txpkt_hdalign() changed the
behavior and now always returns 0. This resulted in a
regression which basically renders the device useless.

Fixes: 270a6c1f

 ("brcmfmac: rework headroom check in .start_xmit()")
Reported-by: S. Gilles <sgilles@math.umd.edu>
Tested-by: S. Gilles <sgilles@math.umd.edu>
Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Pull networking fixes from David Miller:

 1) BPF verifier signed/unsigned value tracking fix, from Daniel
    Borkmann, Edward Cree, and Josef Bacik.

 2) Fix memory allocation length when setting up calls to
    ->ndo_set_mac_address, from Cong Wang.

 3) Add a new cxgb4 device ID, from Ganesh Goudar.

 4) Fix FIB refcount handling, we have to set it's initial value before
    the configure callback (which can bump it). From David Ahern.

 5) Fix double-free in qcom/emac driver, from Timur Tabi.

 6) A bunch of gcc-7 string format overflow warning fixes from Arnd
    Bergmann.

 7) Fix link level headroom tests in ip_do_fragment(), from Vasily
    Averin.

 8) Fix chunk walking in SCTP when iterating over error and parameter
    headers. From Alexander Potapenko.

 9) TCP BBR congestion control fixes from Neal Cardwell.

10) Fix SKB fragment handling in bcmgenet driver, from Doug Berger.

11) BPF_CGROUP_RUN_PROG_SOCK_OPS needs to check for null __sk, from Cong
    Wang.

12) xmit_recursion in ppp driver needs to be per-device not per-cpu,
    from Gao Feng.

13) Cannot release skb->dst in UDP if IP options processing needs it.
    From Paolo Abeni.

14) Some netdev ioctl ifr_name[] NULL termination fixes. From Alexander
    Levin and myself.

15) Revert some rtnetlink notification changes that are causing
    regressions, from David Ahern.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (83 commits)
  net: bonding: Fix transmit load balancing in balance-alb mode
  rds: Make sure updates to cp_send_gen can be observed
  net: ethernet: ti: cpsw: Push the request_irq function to the end of probe
  ipv4: initialize fib_trie prior to register_netdev_notifier call.
  rtnetlink: allocate more memory for dev_set_mac_address()
  net: dsa: b53: Add missing ARL entries for BCM53125
  bpf: more tests for mixed signed and unsigned bounds checks
  bpf: add test for mixed signed and unsigned bounds checks
  bpf: fix up test cases with mixed signed/unsigned bounds
  bpf: allow to specify log level and reduce it for test_verifier
  bpf: fix mixed signed/unsigned derived min/max value bounds
  ipv6: avoid overflow of offset in ip6_find_1stfragopt
  net: tehuti: don't process data if it has not been copied from userspace
  Revert "rtnetlink: Do not generate notifications for CHANGEADDR event"
  net: dsa: mv88e6xxx: Enable CMODE config support for 6390X
  dt-binding: ptp: Add SoC compatibility strings for dte ptp clock
  NET: dwmac: Make dwmac reset unconditional
  net: Zero terminate ifr_name in dev_ifname().
  wireless: wext: terminate ifr name coming from userspace
  netfilter: fix netfilter_net_init() return
  ...

balance-alb mode used to have transmit dynamic load balancing feature
enabled by default.  However, transmit dynamic load balancing no longer
works in balance-alb after commit 8b426dc5 ("bonding: remove
hardcoded value").

Both balance-tlb and balance-alb use the function bond_do_alb_xmit() to
send packets.  This function uses the parameter tlb_dynamic_lb.
tlb_dynamic_lb used to have the default value of 1 for balance-alb, but
now the value is set to 0 except in balance-tlb.

Re-enable transmit dyanmic load balancing by initializing tlb_dynamic_lb
for balance-alb similar to balance-tlb.

Fixes: 8b426dc5

 ("bonding: remove hardcoded value")
Signed-off-by: Kosuke Tatsukawa <tatsu@ab.jp.nec.com>
Acked-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

cp->cp_send_gen is treated as a normal variable, although it may be
used by different threads.

This is fixed by using {READ,WRITE}_ONCE when it is incremented and
READ_ONCE when it is read outside the {acquire,release}_in_xmit
protection.

Normative reference from the Linux-Kernel Memory Model:

    Loads from and stores to shared (but non-atomic) variables should
    be protected with the READ_ONCE(), WRITE_ONCE(), and
    ACCESS_ONCE().

Clause 5.1.2.4/25 in the C standard is also relevant.

Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com>
Reviewed-by: Knut Omang <knut.omang@oracle.com>
Acked-by: Santosh Shilimkar <santosh.shilimkar@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Push the request_irq function to the end of probe so as
to ensure all the required fields are populated in the event
of an ISR getting executed right after requesting the irq.

Currently while loading the crash kernel a crash was seen as
soon as devm_request_threaded_irq was called. This was due to
n->poll being NULL which is called as part of net_rx_action
function.

Suggested-by: Sekhar Nori <nsekhar@ti.com>
Signed-off-by: Keerthy <j-keerthy@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Net stack initialization currently initializes fib-trie after the
first call to netdevice_notifier() call. In fact fib_trie initialization
needs to happen before first rtnl_register(). It does not cause any problem
since there are no devices UP at this moment, but trying to bring 'lo'
UP at initialization would make this assumption wrong and exposes the issue.

Fixes following crash

 Call Trace:
  ? alternate_node_alloc+0x76/0xa0
  fib_table_insert+0x1b7/0x4b0
  fib_magic.isra.17+0xea/0x120
  fib_add_ifaddr+0x7b/0x190
  fib_netdev_event+0xc0/0x130
  register_netdevice_notifier+0x1c1/0x1d0
  ip_fib_init+0x72/0x85
  ip_rt_init+0x187/0x1e9
  ip_init+0xe/0x1a
  inet_init+0x171/0x26c
  ? ipv4_offload_init+0x66/0x66
  do_one_initcall+0x43/0x160
  kernel_init_freeable+0x191/0x219
  ? rest_init+0x80/0x80
  kernel_init+0xe/0x150
  ret_from_fork+0x22/0x30
 Code: f6 46 23 04 74 86 4c 89 f7 e8 ae 45 01 00 49 89 c7 4d 85 ff 0f 85 7b ff ff ff 31 db eb 08 4c 89 ff e8 16 47 01 00 48 8b 44 24 38 <45> 8b 6e 14 4d 63 76 74 48 89 04 24 0f 1f 44 00 00 48 83 c4 08
 RIP: kmem_cache_alloc+0xcf/0x1c0 RSP: ffff9b1500017c28
 CR2: 0000000000000014

Fixes: 7b1a74fd ("[NETNS]: Refactor fib initialization so it can handle multiple namespaces.")
Fixes: 7f9b8052

 ("[IPV4]: fib hash|trie initialization")

Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

virtnet_set_mac_address() interprets mac address as struct
sockaddr, but upper layer only allocates dev->addr_len
which is ETH_ALEN + sizeof(sa_family_t) in this case.

We lack a unified definition for mac address, so just fix
the upper layer, this also allows drivers to interpret it
to struct sockaddr freely.

Reported-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The BCM53125 entry was missing an arl_entries member which would
basically prevent the ARL search from terminating properly. This switch
has 4 ARL entries, so add that.

Fixes: 1da6df85

 ("net: dsa: b53: Implement ARL add/del/dump operations")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
BPF map value adjust fix

First patch in the series is the actual fix and the remaining
patches are just updates to selftests.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Add a couple of more test cases to BPF selftests that are related
to mixed signed and unsigned checks.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

These failed due to a bug in verifier bounds handling.

Signed-off-by: Edward Cree <ecree@solarflare.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Fix the few existing test cases that used mixed signed/unsigned
bounds and switch them only to one flavor. Reason why we need this
is that proper boundaries cannot be derived from mixed tests.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

For the test_verifier case, it's quite hard to parse log level 2 to
figure out what's causing an issue when used to log level 1. We do
want to use bpf_verify_program() in order to simulate some of the
tests with strict alignment. So just add an argument to pass the level
and put it to 1 for test_verifier.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Edward reported that there's an issue in min/max value bounds
tracking when signed and unsigned compares both provide hints
on limits when having unknown variables. E.g. a program such
as the following should have been rejected:

   0: (7a) *(u64 *)(r10 -8) = 0
   1: (bf) r2 = r10
   2: (07) r2 += -8
   3: (18) r1 = 0xffff8a94cda93400
   5: (85) call bpf_map_lookup_elem#1
   6: (15) if r0 == 0x0 goto pc+7
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R10=fp
   7: (7a) *(u64 *)(r10 -16) = -8
   8: (79) r1 = *(u64 *)(r10 -16)
   9: (b7) r2 = -1
  10: (2d) if r1 > r2 goto pc+3
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R1=inv,min_value=0
  R2=imm-1,max_value=18446744073709551615,min_align=1 R10=fp
  11: (65) if r1 s> 0x1 goto pc+2
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R1=inv,min_value=0,max_value=1
  R2=imm-1,max_value=18446744073709551615,min_align=1 R10=fp
  12: (0f) r0 += r1
  13: (72) *(u8 *)(r0 +0) = 0
  R0=map_value_adj(ks=8,vs=8,id=0),min_value=0,max_value=1 R1=inv,min_value=0,max_value=1
  R2=imm-1,max_value=18446744073709551615,min_align=1 R10=fp
  14: (b7) r0 = 0
  15: (95) exit

What happens is that in the first part ...

   8: (79) r1 = *(u64 *)(r10 -16)
   9: (b7) r2 = -1
  10: (2d) if r1 > r2 goto pc+3

... r1 carries an unsigned value, and is compared as unsigned
against a register carrying an immediate. Verifier deduces in
reg_set_min_max() that since the compare is unsigned and operation
is greater than (>), that in the fall-through/false case, r1's
minimum bound must be 0 and maximum bound must be r2. Latter is
larger than the bound and thus max value is reset back to being
'invalid' aka BPF_REGISTER_MAX_RANGE. Thus, r1 state is now
'R1=inv,min_value=0'. The subsequent test ...

  11: (65) if r1 s> 0x1 goto pc+2

... is a signed compare of r1 with immediate value 1. Here,
verifier deduces in reg_set_min_max() that since the compare
is signed this time and operation is greater than (>), that
in the fall-through/false case, we can deduce that r1's maximum
bound must be 1, meaning with prior test, we result in r1 having
the following state: R1=inv,min_value=0,max_value=1. Given that
the actual value this holds is -8, the bounds are wrongly deduced.
When this is being added to r0 which holds the map_value(_adj)
type, then subsequent store access in above case will go through
check_mem_access() which invokes check_map_access_adj(), that
will then probe whether the map memory is in bounds based
on the min_value and max_value as well as access size since
the actual unknown value is min_value <= x <= max_value; commit
fce366a9 ("bpf, verifier: fix alu ops against map_value{,
_adj} register types") provides some more explanation on the
semantics.

It's worth to note in this context that in the current code,
min_value and max_value tracking are used for two things, i)
dynamic map value access via check_map_access_adj() and since
commit 06c1c049 ("bpf: allow helpers access to variable memory")
ii) also enforced at check_helper_mem_access() when passing a
memory address (pointer to packet, map value, stack) and length
pair to a helper and the length in this case is an unknown value
defining an access range through min_value/max_value in that
case. The min_value/max_value tracking is /not/ used in the
direct packet access case to track ranges. However, the issue
also affects case ii), for example, the following crafted program
based on the same principle must be rejected as well:

   0: (b7) r2 = 0
   1: (bf) r3 = r10
   2: (07) r3 += -512
   3: (7a) *(u64 *)(r10 -16) = -8
   4: (79) r4 = *(u64 *)(r10 -16)
   5: (b7) r6 = -1
   6: (2d) if r4 > r6 goto pc+5
  R1=ctx R2=imm0,min_value=0,max_value=0,min_align=2147483648 R3=fp-512
  R4=inv,min_value=0 R6=imm-1,max_value=18446744073709551615,min_align=1 R10=fp
   7: (65) if r4 s> 0x1 goto pc+4
  R1=ctx R2=imm0,min_value=0,max_value=0,min_align=2147483648 R3=fp-512
  R4=inv,min_value=0,max_value=1 R6=imm-1,max_value=18446744073709551615,min_align=1
  R10=fp
   8: (07) r4 += 1
   9: (b7) r5 = 0
  10: (6a) *(u16 *)(r10 -512) = 0
  11: (85) call bpf_skb_load_bytes#26
  12: (b7) r0 = 0
  13: (95) exit

Meaning, while we initialize the max_value stack slot that the
verifier thinks we access in the [1,2] range, in reality we
pass -7 as length which is interpreted as u32 in the helper.
Thus, this issue is relevant also for the case of helper ranges.
Resetting both bounds in check_reg_overflow() in case only one
of them exceeds limits is also not enough as similar test can be
created that uses values which are within range, thus also here
learned min value in r1 is incorrect when mixed with later signed
test to create a range:

   0: (7a) *(u64 *)(r10 -8) = 0
   1: (bf) r2 = r10
   2: (07) r2 += -8
   3: (18) r1 = 0xffff880ad081fa00
   5: (85) call bpf_map_lookup_elem#1
   6: (15) if r0 == 0x0 goto pc+7
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R10=fp
   7: (7a) *(u64 *)(r10 -16) = -8
   8: (79) r1 = *(u64 *)(r10 -16)
   9: (b7) r2 = 2
  10: (3d) if r2 >= r1 goto pc+3
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R1=inv,min_value=3
  R2=imm2,min_value=2,max_value=2,min_align=2 R10=fp
  11: (65) if r1 s> 0x4 goto pc+2
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0
  R1=inv,min_value=3,max_value=4 R2=imm2,min_value=2,max_value=2,min_align=2 R10=fp
  12: (0f) r0 += r1
  13: (72) *(u8 *)(r0 +0) = 0
  R0=map_value_adj(ks=8,vs=8,id=0),min_value=3,max_value=4
  R1=inv,min_value=3,max_value=4 R2=imm2,min_value=2,max_value=2,min_align=2 R10=fp
  14: (b7) r0 = 0
  15: (95) exit

This leaves us with two options for fixing this: i) to invalidate
all prior learned information once we switch signed context, ii)
to track min/max signed and unsigned boundaries separately as
done in [0]. (Given latter introduces major changes throughout
the whole verifier, it's rather net-next material, thus this
patch follows option i), meaning we can derive bounds either
from only signed tests or only unsigned tests.) There is still the
case of adjust_reg_min_max_vals(), where we adjust bounds on ALU
operations, meaning programs like the following where boundaries
on the reg get mixed in context later on when bounds are merged
on the dst reg must get rejected, too:

   0: (7a) *(u64 *)(r10 -8) = 0
   1: (bf) r2 = r10
   2: (07) r2 += -8
   3: (18) r1 = 0xffff89b2bf87ce00
   5: (85) call bpf_map_lookup_elem#1
   6: (15) if r0 == 0x0 goto pc+6
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R10=fp
   7: (7a) *(u64 *)(r10 -16) = -8
   8: (79) r1 = *(u64 *)(r10 -16)
   9: (b7) r2 = 2
  10: (3d) if r2 >= r1 goto pc+2
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R1=inv,min_value=3
  R2=imm2,min_value=2,max_value=2,min_align=2 R10=fp
  11: (b7) r7 = 1
  12: (65) if r7 s> 0x0 goto pc+2
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R1=inv,min_value=3
  R2=imm2,min_value=2,max_value=2,min_align=2 R7=imm1,max_value=0 R10=fp
  13: (b7) r0 = 0
  14: (95) exit

  from 12 to 15: R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0
  R1=inv,min_value=3 R2=imm2,min_value=2,max_value=2,min_align=2 R7=imm1,min_value=1 R10=fp
  15: (0f) r7 += r1
  16: (65) if r7 s> 0x4 goto pc+2
  R0=map_value(ks=8,vs=8,id=0),min_value=0,max_value=0 R1=inv,min_value=3
  R2=imm2,min_value=2,max_value=2,min_align=2 R7=inv,min_value=4,max_value=4 R10=fp
  17: (0f) r0 += r7
  18: (72) *(u8 *)(r0 +0) = 0
  R0=map_value_adj(ks=8,vs=8,id=0),min_value=4,max_value=4 R1=inv,min_value=3
  R2=imm2,min_value=2,max_value=2,min_align=2 R7=inv,min_value=4,max_value=4 R10=fp
  19: (b7) r0 = 0
  20: (95) exit

Meaning, in adjust_reg_min_max_vals() we must also reset range
values on the dst when src/dst registers have mixed signed/
unsigned derived min/max value bounds with one unbounded value
as otherwise they can be added together deducing false boundaries.
Once both boundaries are established from either ALU ops or
compare operations w/o mixing signed/unsigned insns, then they
can safely be added to other regs also having both boundaries
established. Adding regs with one unbounded side to a map value
where the bounded side has been learned w/o mixing ops is
possible, but the resulting map value won't recover from that,
meaning such op is considered invalid on the time of actual
access. Invalid bounds are set on the dst reg in case i) src reg,
or ii) in case dst reg already had them. The only way to recover
would be to perform i) ALU ops but only 'add' is allowed on map
value types or ii) comparisons, but these are disallowed on
pointers in case they span a range. This is fine as only BPF_JEQ
and BPF_JNE may be performed on PTR_TO_MAP_VALUE_OR_NULL registers
which potentially turn them into PTR_TO_MAP_VALUE type depending
on the branch, so only here min/max value cannot be invalidated
for them.

In terms of state pruning, value_from_signed is considered
as well in states_equal() when dealing with adjusted map values.
With regards to breaking existing programs, there is a small
risk, but use-cases are rather quite narrow where this could
occur and mixing compares probably unlikely.

Joint work with Josef and Edward.

  [0] https://lists.iovisor.org/pipermail/iovisor-dev/2017-June/000822.html

Fixes: 48461135

 ("bpf: allow access into map value arrays")
Reported-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>