Linus Torvalds [Tue, 19 Dec 2023 00:47:21 +0000 (16:47 -0800)]
Merge tag 'hid-for-linus-2023121901' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid
Pull HID fixes from Jiri Kosina:
- fix for division by zero in Nintendo driver when generic joycon is
attached, reported and fixed by SteamOS folks (Guilherme G. Piccoli)
- GCC-7 build fix (which is a good cleanup anyway) for Nintendo driver
(Ryan McClelland)
* tag 'hid-for-linus-2023121901' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid:
HID: nintendo: Prevent divide-by-zero on code
HID: nintendo: fix initializer element is not constant error
This PR is larger than usual and contains changes in various parts
of the kernel.
The main changes are:
1) Fix kCFI bugs in BPF, from Peter Zijlstra.
End result: all forms of indirect calls from BPF into kernel
and from kernel into BPF work with CFI enabled. This allows BPF
to work with CONFIG_FINEIBT=y.
2) Introduce BPF token object, from Andrii Nakryiko.
It adds an ability to delegate a subset of BPF features from privileged
daemon (e.g., systemd) through special mount options for userns-bound
BPF FS to a trusted unprivileged application. The design accommodates
suggestions from Christian Brauner and Paul Moore.
3) Various verifier improvements and fixes, from Andrii Nakryiko, Andrei Matei.
- Complete precision tracking support for register spills
- Fix verification of possibly-zero-sized stack accesses
- Fix access to uninit stack slots
- Track aligned STACK_ZERO cases as imprecise spilled registers.
It improves the verifier "instructions processed" metric from single
digit to 50-60% for some programs.
- Fix verifier retval logic
4) Support for VLAN tag in XDP hints, from Larysa Zaremba.
5) Allocate BPF trampoline via bpf_prog_pack mechanism, from Song Liu.
End result: better memory utilization and lower I$ miss for calls to BPF
via BPF trampoline.
6) Fix race between BPF prog accessing inner map and parallel delete,
from Hou Tao.
7) Add bpf_xdp_get_xfrm_state() kfunc, from Daniel Xu.
It allows BPF interact with IPSEC infra. The intent is to support
software RSS (via XDP) for the upcoming ipsec pcpu work.
Experiments on AWS demonstrate single tunnel pcpu ipsec reaching
line rate on 100G ENA nics.
8) Expand bpf_cgrp_storage to support cgroup1 non-attach, from Yafang Shao.
9) BPF file verification via fsverity, from Song Liu.
It allows BPF progs get fsverity digest.
* tag 'for-netdev' of https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next: (164 commits)
bpf: Ensure precise is reset to false in __mark_reg_const_zero()
selftests/bpf: Add more uprobe multi fail tests
bpf: Fail uprobe multi link with negative offset
selftests/bpf: Test the release of map btf
s390/bpf: Fix indirect trampoline generation
selftests/bpf: Temporarily disable dummy_struct_ops test on s390
x86/cfi,bpf: Fix bpf_exception_cb() signature
bpf: Fix dtor CFI
cfi: Add CFI_NOSEAL()
x86/cfi,bpf: Fix bpf_struct_ops CFI
x86/cfi,bpf: Fix bpf_callback_t CFI
x86/cfi,bpf: Fix BPF JIT call
cfi: Flip headers
selftests/bpf: Add test for abnormal cnt during multi-kprobe attachment
selftests/bpf: Don't use libbpf_get_error() in kprobe_multi_test
selftests/bpf: Add test for abnormal cnt during multi-uprobe attachment
bpf: Limit the number of kprobes when attaching program to multiple kprobes
bpf: Limit the number of uprobes when attaching program to multiple uprobes
bpf: xdp: Register generic_kfunc_set with XDP programs
selftests/bpf: utilize string values for delegate_xxx mount options
...
====================
Jakub Kicinski [Tue, 19 Dec 2023 00:17:33 +0000 (16:17 -0800)]
Merge tag 'wireless-next-2023-12-18' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next
Kalle Valo says:
====================
wireless-next patches for v6.8
The second features pull request for v6.8. A bigger one this time with
changes both to stack and drivers. We have a new Wifi band RFI (WBRF)
mitigation feature for which we pulled an immutable branch shared with
other subsystems. And, as always, other new features and bug fixes all
over.
Major changes:
cfg80211/mac80211
* AMD ACPI based Wifi band RFI (WBRF) mitigation feature
* Basic Service Set (BSS) usage reporting
* TID to link mapping support
* mac80211 hardware flag to disallow puncturing
iwlwifi
* new debugfs file fw_dbg_clear
mt76
* NVMEM EEPROM improvements
* mt7996 Extremely High Throughpu (EHT) improvements
* mt7996 Wireless Ethernet Dispatcher (WED) support
* mt7996 36-bit DMA support
ath12k
* support one MSI vector
* WCN7850: support AP mode
* tag 'wireless-next-2023-12-18' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next: (207 commits)
wifi: mt76: mt7996: Use DECLARE_FLEX_ARRAY() and fix -Warray-bounds warnings
wifi: ath11k: workaround too long expansion sparse warnings
Revert "wifi: ath12k: use ATH12K_PCI_IRQ_DP_OFFSET for DP IRQ"
wifi: rt2x00: remove useless code in rt2x00queue_create_tx_descriptor()
wifi: rtw89: only reset BB/RF for existing WiFi 6 chips while starting up
wifi: rtw89: add DBCC H2C to notify firmware the status
wifi: rtw89: mac: add suffix _ax to MAC functions
wifi: rtw89: mac: add flags to check if CMAC and DMAC are enabled
wifi: rtw89: 8922a: add power on/off functions
wifi: rtw89: add XTAL SI for WiFi 7 chips
wifi: rtw89: phy: print out RFK log with formatted string
wifi: rtw89: parse and print out RFK log from C2H events
wifi: rtw89: add C2H event handlers of RFK log and report
wifi: rtw89: load RFK log format string from firmware file
wifi: rtw89: fw: add version field to BB MCU firmware element
wifi: rtw89: fw: load TX power track tables from fw_element
wifi: mwifiex: configure BSSID consistently when starting AP
wifi: mwifiex: add extra delay for firmware ready
wifi: mac80211: sta_info.c: fix sentence grammar
wifi: mac80211: rx.c: fix sentence grammar
...
====================
Andrii Nakryiko [Mon, 18 Dec 2023 17:36:01 +0000 (09:36 -0800)]
bpf: Ensure precise is reset to false in __mark_reg_const_zero()
It is safe to always start with imprecise SCALAR_VALUE register.
Previously __mark_reg_const_zero() relied on caller to reset precise
mark, but it's very error prone and we already missed it in a few
places. So instead make __mark_reg_const_zero() reset precision always,
as it's a safe default for SCALAR_VALUE. Explanation is basically the
same as for why we are resetting (or rather not setting) precision in
current state. If necessary, precision propagation will set it to
precise correctly.
As such, also remove a big comment about forward precision propagation
in mark_reg_stack_read() and avoid unnecessarily setting precision to
true after reading from STACK_ZERO stack. Again, precision propagation
will correctly handle this, if that SCALAR_VALUE register will ever be
needed to be precise.
====================
tools/net/ynl: Add 'sub-message' support to ynl
This patchset adds a 'sub-message' attribute type to the netlink-raw
schema and implements it in ynl. This provides support for kind-specific
options attributes as used in rt_link and tc raw netlink families.
A description of the new 'sub-message' attribute type and the
corresponding sub-message definitions is provided in patch 3.
The patchset includes updates to the rt_link spec and a new tc spec that
make use of the new 'sub-message' attribute type.
As mentioned in patch 4, encode support is not yet implemented in ynl
and support for sub-message selectors at a different nest level from the
key attribute is not yet supported. I plan to work on these in follow-up
patches.
Patches 1 is code cleanup in ynl
Patches 2-4 add sub-message support to the schema and ynl with
documentation updates.
Patch 5 adds binary and pad support to structs in netlink-raw.
Patches 6-8 contain specs that use the sub-message attribute type.
Patches 9-13 update ynl-gen-rst and its make target
====================
Donald Hunter [Fri, 15 Dec 2023 09:37:19 +0000 (09:37 +0000)]
tools/net/ynl-gen-rst: Remove bold from attribute-set headings
The generated .rst for attribute-sets currently uses a sub-sub-heading
for each attribute, with the attribute name in bold. This makes
attributes stand out more than the attribute-set sub-headings they are
part of.
Remove the bold markup from attribute sub-sub-headings.
Donald Hunter [Fri, 15 Dec 2023 09:37:16 +0000 (09:37 +0000)]
doc/netlink: Regenerate netlink .rst files if ynl-gen-rst changes
Add ynl-gen-rst.py to the dependencies for the netlink .rst files in the
doc Makefile so that the docs get regenerated if the ynl-gen-rst.py
script is modified. Use $(Q) to honour V=1 in the rules that run
ynl-gen-rst.py
Donald Hunter [Fri, 15 Dec 2023 09:37:14 +0000 (09:37 +0000)]
doc/netlink/specs: use pad in structs in rt_link
The rt_link spec was using pad1, pad2 attributes in structs which
appears in the ynl output. Replace this with the 'pad' type which
doesn't pollute the output.
Donald Hunter [Fri, 15 Dec 2023 09:37:09 +0000 (09:37 +0000)]
doc/netlink: Add sub-message support to netlink-raw
Add a 'sub-message' attribute type with a selector that supports
polymorphic attribute formats for raw netlink families like tc.
A sub-message attribute uses the value of another attribute as a
selector key to choose the right sub-message format. For example if the
following attribute has already been decoded:
{ "kind": "gre" }
and we encounter the following attribute spec:
-
name: data
type: sub-message
sub-message: linkinfo-data-msg
selector: kind
Then we look for a sub-message definition called 'linkinfo-data-msg' and
use the value of the 'kind' attribute i.e. 'gre' as the key to choose
the correct format for the sub-message:
This would decode the attribute value as a sub-message with the
attribute-set called 'linkinfo-gre-attrs' as the attribute space.
A sub-message can have an optional 'fixed-header' followed by zero or
more attributes from an attribute-set. For example the following
'tc-options-msg' sub-message defines message formats that use a mixture
of fixed-header, attribute-set or both together:
Guillaume says:
> I believe commit 5f7fc5d69f6e ("SUNRPC: Resupply rq_pages from
> node-local memory") in Linux 6.5+ is incorrect. It passes
> unconditionally rq_pool->sp_id as the NUMA node.
>
> While the comment in the svc_pool declaration in sunrpc/svc.h says
> that sp_id is also the NUMA node id, it might not be the case if
> the svc is created using svc_create_pooled(). svc_created_pooled()
> can use the per-cpu pool mode therefore in this case sp_id would
> be the cpu id.
Fix this by reverting now. At a later point this minor optimization,
and the deceptive labeling of the sp_id field, can be revisited.
Since it's a divide-by-0 error, by tracking the code for potential
denominator issues, we've spotted 2 places in which this could happen;
so let's guard against the possibility and log in the kernel if the
condition happens. This is specially useful since some data that
fills some denominators are read from the joycon HW in some cases,
increasing the potential for flaws.
idpf: refactor some missing field get/prep conversions
Most of idpf correctly uses FIELD_GET and FIELD_PREP, but a couple spots
were missed so fix those.
Automated conversion with coccinelle script and manually fixed up,
including audits for opportunities to convert to {get,encode,replace}
bits functions.
Add conversions to le16_get/encode/replace_bits where appropriate. And
in one place fix up a cast from a u16 to a u16.
@prep2@
constant shift,mask;
type T;
expression a;
@@
-(((T)(a) << shift) & mask)
+FIELD_PREP(mask, a)
@prep@
constant shift,mask;
type T;
expression a;
@@
-((T)((a) << shift) & mask)
+FIELD_PREP(mask, a)
@get@
constant shift,mask;
type T;
expression a;
@@
-((T)((a) & mask) >> shift)
+FIELD_GET(mask, a)
and applied via:
spatch --sp-file field_prep.cocci --in-place --dir \
drivers/net/ethernet/intel/
It was found while doing further testing of the previous commit fbf32a9bab91 ("ice: field get conversion") that one of the FIELD_GET
conversions should really be a FIELD_PREP. The previous code was styled
as a match to the FIELD_GET conversion, which always worked because the
shift value was 0. The code makes way more sense as a FIELD_PREP and
was in fact the only FIELD_GET with two constant arguments in this
series.
Didn't squash this patch to make it easier to call out the
(non-impactful) bug.
Refactor the i40e driver to use FIELD_GET() for mask and shift reads,
which reduces lines of code and adds clarity of intent.
This code was generated by the following coccinelle/spatch script and
then manually repaired.
While making one of the conversions, an if() check was inverted to
return early and avoid un-necessary indentation of the remainder of the
function. In some other cases a stack variable was moved inside the
block where it was used while doing cleanups/review.
A couple places were changed to use le16_get_bits() instead of FIELD_GET
with a le16_to_cpu combination.
@get@
constant shift,mask;
metavariable type T;
expression a;
@@
-(((T)(a) & mask) >> shift)
+FIELD_GET(mask, a)
and applied via:
spatch --sp-file field_prep.cocci --in-place --dir \
drivers/net/ethernet/intel/
While converting to FIELD_PREP() and FIELD_GET(), it was noticed that
some of the RSS defines had *included* the shift in their definitions.
This is completely outside of normal, such that a developer could easily
make a mistake and shift at the usage site (like when using
FIELD_PREP()).
Rename the defines and set them to the "pre-shifted values" so they
match the template the driver normally uses for masks and the member
bits of the mask, which also allows the driver to use FIELD_PREP
correctly with these values. Use GENMASK() for this changed MASK value.
Refactor ice driver to use FIELD_PREP(), which reduces lines of code
and adds clarity of intent.
This code was generated by the following coccinelle/spatch script and
then manually repaired.
Several places I changed to OR into a single variable with |= instead of
using a multi-line statement with trailing OR operators, as it
(subjectively) makes the code clearer.
A local variable vmvf_and_timeout was created and used to avoid multiple
logical ORs being __le16 converted, which shortened some lines and makes
the code cleaner.
Also clean up a couple of places where conversions were made to have the
code read more clearly/consistently.
@prep2@
constant shift,mask;
type T;
expression a;
@@
-(((T)(a) << shift) & mask)
+FIELD_PREP(mask, a)
@prep@
constant shift,mask;
type T;
expression a;
@@
-((T)((a) << shift) & mask)
+FIELD_PREP(mask, a)
Refactor iavf driver to use FIELD_PREP(), which reduces lines of code
and adds clarity of intent.
This code was generated by the following coccinelle/spatch script and
then manually repaired.
Clean up a couple spots in the code that had repetitive
y = cpu_to_*((blah << blah_blah) & blat)
y |= cpu_to_*((blahs << blahs_blahs) & blats)
to
x = FIELD_PREP(blat blah)
x |= FIELD_PREP(blats, blahs)
y = cpu_to_*(x);
@prep2@
constant shift,mask;
type T;
expression a;
@@
-(((T)(a) << shift) & mask)
+FIELD_PREP(mask, a)
@prep@
constant shift,mask;
type T;
expression a;
@@
-((T)((a) << shift) & mask)
+FIELD_PREP(mask, a)
This series is introducing the use of FIELD_GET and FIELD_PREP which
requires bitfield.h to be included. Fix all the includes in this one
change, and rearrange includes into alphabetical order to ease
readability and future maintenance.
virtchnl.h and it's usage was modified to have it's own includes as it
should. This required including bits.h for virtchnl.h.
For more than 15 years this code has passed in a request for a page and
masked off that page when read/writing. This code has been here forever,
but FIELD_PREP finds the bug when converted to use it. Change the code
to do exactly the same thing but allow the conversion to FIELD_PREP in a
later patch. To make it clear what we lost when making this change I
left a comment, but there is no point to change the code to generate a
correct sequence at this point.
This is not a Fixes tagged patch on purpose because it doesn't change
the binary output.
Linus Torvalds [Mon, 18 Dec 2023 19:11:09 +0000 (11:11 -0800)]
Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Pull SCSI fixes from James Bottomley:
"Two medium sized fixes, both in drivers.
The UFS one adds parsing of clock info structures, which is required
by some host drivers and the aacraid one reverts the IRQ affinity
mapping patch which has been causing regressions noted in kernel
bugzilla 217599"
* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
scsi: ufs: core: Store min and max clk freq from OPP table
Revert "scsi: aacraid: Reply queue mapping to CPUs based on IRQ affinity"
Linus Torvalds [Mon, 18 Dec 2023 18:59:57 +0000 (10:59 -0800)]
Merge tag 'spi-fix-v6.7-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi
Pull spi fixes from Mark Brown:
"A few bigger things here, the main one being that there were changes
to the atmel driver in this cycle which made it possible to kill
transfers being used for filesystem I/O which turned out to be very
disruptive, the series of patches here undoes that and hardens things
up further.
There's also a few smaller driver specific changes, the main one being
to revert a change that duplicted delays"
* tag 'spi-fix-v6.7-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
spi: atmel: Fix clock issue when using devices with different polarities
spi: spi-imx: correctly configure burst length when using dma
spi: cadence: revert "Add SPI transfer delays"
spi: atmel: Prevent spi transfers from being killed
spi: atmel: Drop unused defines
spi: atmel: Do not cancel a transfer upon any signal
Mikulas Patocka [Tue, 5 Dec 2023 15:39:16 +0000 (16:39 +0100)]
dm-integrity: don't modify bio's immutable bio_vec in integrity_metadata()
__bio_for_each_segment assumes that the first struct bio_vec argument
doesn't change - it calls "bio_advance_iter_single((bio), &(iter),
(bvl).bv_len)" to advance the iterator. Unfortunately, the dm-integrity
code changes the bio_vec with "bv.bv_len -= pos". When this code path
is taken, the iterator would be out of sync and dm-integrity would
report errors. This happens if the machine is out of memory and
"kmalloc" fails.
Fix this bug by making a copy of "bv" and changing the copy instead.
Yu Kuai [Fri, 24 Nov 2023 07:59:53 +0000 (15:59 +0800)]
dm-raid: delay flushing event_work() after reconfig_mutex is released
After commit db5e653d7c9f ("md: delay choosing sync action to
md_start_sync()"), md_start_sync() will hold 'reconfig_mutex', however,
in order to make sure event_work is done, __md_stop() will flush
workqueue with reconfig_mutex grabbed, hence if sync_work is still
pending, deadlock will be triggered.
Fortunately, former pacthes to fix stopping sync_thread already make sure
all sync_work is done already, hence such deadlock is not possible
anymore. However, in order not to cause confusions for people by this
implicit dependency, delay flushing event_work to dm-raid where
'reconfig_mutex' is not held, and add some comments to emphasize that
the workqueue can't be flushed with 'reconfig_mutex'.
Jiri Olsa [Sun, 17 Dec 2023 21:55:37 +0000 (22:55 +0100)]
bpf: Fail uprobe multi link with negative offset
Currently the __uprobe_register will return 0 (success) when called with
negative offset. The reason is that the call to register_for_each_vma and
then build_map_info won't return error for negative offset. They just won't
do anything - no matching vma is found so there's no registered breakpoint
for the uprobe.
I don't think we can change the behaviour of __uprobe_register and fail
for negative uprobe offset, because apps might depend on that already.
But I think we can still make the change and check for it on bpf multi
link syscall level.
Also moving the __get_user call and check for the offsets to the top of
loop, to fail early without extra __get_user calls for ref_ctr_offset
and cookie arrays.
Larysa Zaremba [Tue, 12 Dec 2023 09:29:01 +0000 (10:29 +0100)]
ice: Fix PF with enabled XDP going no-carrier after reset
Commit 6624e780a577fc596788 ("ice: split ice_vsi_setup into smaller
functions") has refactored a bunch of code involved in PFR. In this
process, TC queue number adjustment for XDP was lost. Bring it back.
Lack of such adjustment causes interface to go into no-carrier after a
reset, if XDP program is attached, with the following message:
ice 0000:b1:00.0: Failed to set LAN Tx queue context, error: -22
ice 0000:b1:00.0 ens801f0np0: Failed to open VSI 0x0006 on switch 0x0001
ice 0000:b1:00.0: enable VSI failed, err -22, VSI index 0, type ICE_VSI_PF
ice 0000:b1:00.0: PF VSI rebuild failed: -22
ice 0000:b1:00.0: Rebuild failed, unload and reload driver
Fixes: 6624e780a577 ("ice: split ice_vsi_setup into smaller functions") Reviewed-by: Przemek Kitszel <[email protected]> Signed-off-by: Larysa Zaremba <[email protected]> Reviewed-by: Simon Horman <[email protected]> Tested-by: Chandan Kumar Rout <[email protected]> (A Contingent Worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
Dave Ertman [Mon, 11 Dec 2023 21:19:28 +0000 (13:19 -0800)]
ice: alter feature support check for SRIOV and LAG
Previously, the ice driver had support for using a handler for bonding
netdev events to ensure that conflicting features were not allowed to be
activated at the same time. While this was still in place, additional
support was added to specifically support SRIOV and LAG together. These
both utilized the netdev event handler, but the SRIOV and LAG feature was
behind a capabilities feature check to make sure the current NVM has
support.
The exclusion part of the event handler should be removed since there are
users who have custom made solutions that depend on the non-exclusion of
features.
Wrap the creation/registration and cleanup of the event handler and
associated structs in the probe flow with a feature check so that the
only systems that support the full implementation of LAG features will
initialize support. This will leave other systems unhindered with
functionality as it existed before any LAG code was added.
Fixes: bb52f42acef6 ("ice: Add driver support for firmware changes for LAG") Reviewed-by: Jesse Brandeburg <[email protected]> Signed-off-by: Dave Ertman <[email protected]> Reviewed-by: Simon Horman <[email protected]> Tested-by: Pucha Himasekhar Reddy <[email protected]> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <[email protected]>
Hou Tao [Sat, 16 Dec 2023 03:55:10 +0000 (11:55 +0800)]
selftests/bpf: Test the release of map btf
When there is bpf_list_head or bpf_rb_root field in map value, the free
of map btf and the free of map value may run concurrently and there may
be use-after-free problem, so add two test cases to demonstrate it. And
the use-after-free problem can been easily reproduced by using bpf_next
tree and a KASAN-enabled kernel.
The first test case tests the racing between the free of map btf and the
free of array map. It constructs the racing by releasing the array map in
the end after other ref-counter of map btf has been released. To delay
the free of array map and make it be invoked after btf_free_rcu() is
invoked, it stresses system_unbound_wq by closing multiple percpu array
maps before it closes the array map.
The second case tests the racing between the free of map btf and the
free of inner map. Beside using the similar method as the first one
does, it uses bpf_map_delete_elem() to delete the inner map and to defer
the release of inner map after one RCU grace period.
The reason for using two skeletons is to prevent the release of outer
map and inner map in map_in_map_btf.c interfering the release of bpf
map in normal_map_btf.c.
Jacob Keller [Wed, 6 Dec 2023 20:19:05 +0000 (12:19 -0800)]
ice: stop trashing VF VSI aggregator node ID information
When creating new VSIs, they are assigned into an aggregator node in the
scheduler tree. Information about which aggregator node a VSI is assigned
into is maintained by the vsi->agg_node structure. In ice_vsi_decfg(), this
information is being destroyed, by overwriting the valid flag and the
agg_id field to zero.
For VF VSIs, this breaks the aggregator node configuration replay, which
depends on this information. This results in VFs being inserted into the
default aggregator node. The resulting configuration will have unexpected
Tx bandwidth sharing behavior.
This was broken by commit 6624e780a577 ("ice: split ice_vsi_setup into
smaller functions"), which added the block to reset the agg_node data.
The vsi->agg_node structure is not managed by the scheduler code, but is
instead a wrapper around an aggregator node ID that is tracked at the VSI
layer. Its been around for a long time, and its primary purpose was for
handling VFs. The SR-IOV VF reset flow does not make use of the standard VSI
rebuild/replay logic, and uses vsi->agg_node as part of its handling to
rebuild the aggregator node configuration.
The logic for aggregator nodes stretches back to early ice driver code from
commit b126bd6bcd67 ("ice: create scheduler aggregator node config and move
VSIs")
The logic in ice_vsi_decfg() which trashes the ice_agg_node data is clearly
wrong. It destroys information that is necessary for handling VF reset,. It
is also not the correct way to actually remove a VSI from an aggregator
node. For that, we need to implement logic in the scheduler code. Further,
non-VF VSIs properly replay their aggregator configuration using existing
scheduler replay logic.
To fix the VF replay logic, remove this broken aggregator node cleanup
logic. This is the simplest way to immediately fix this.
This ensures that VFs will have proper aggregate configuration after a
reset. This is especially important since VFs often perform resets as part
of their reconfiguration flows. Without fixing this, VFs will be placed in
the default aggregator node and Tx bandwidth will not be shared in the
expected and configured manner.
There's nothing wrong with this commit, but this is dead code now
that nothing triggers a CB_GETATTR callback. It can be re-introduced
once the issues with handling conflicting GETATTRs are resolved.
For some reason, the wait_on_bit() in nfsd4_deleg_getattr_conflict()
is waiting forever, preventing a clean server shutdown. The
requesting client might also hang waiting for a reply to the
conflicting GETATTR.
Invoking wait_on_bit() in an nfsd thread context is a hazard. The
correct fix is to replace this wait_on_bit() call site with a
mechanism that defers the conflicting GETATTR until the CB_GETATTR
completes or is known to have failed.
That will require some surgery and extended testing and it's late
in the v6.7-rc cycle, so I'm reverting now in favor of trying again
in a subsequent kernel release.
This is my fault: I should have recognized the ramifications of
calling wait_on_bit() in here before accepting this patch.
platform/x86/amd/pmc: Disable keyboard wakeup on AMD Framework 13
The Laptop 13 (AMD Ryzen 7040Series) BIOS 03.03 has a workaround
included in the EC firmware that will cause the EC to emit a "spurious"
keypress during the resume from s0i3 [1].
This series of keypress events can be observed in the kernel log on
resume.
```
atkbd serio0: Unknown key pressed (translated set 2, code 0x6b on isa0060/serio0).
atkbd serio0: Use 'setkeycodes 6b <keycode>' to make it known.
atkbd serio0: Unknown key released (translated set 2, code 0x6b on isa0060/serio0).
atkbd serio0: Use 'setkeycodes 6b <keycode>' to make it known.
```
In some user flows this is harmless, but if a user has specifically
suspended the laptop and then closed the lid it will cause the laptop
to wakeup. The laptop wakes up because the ACPI SCI triggers when
the lid is closed and when the kernel sees that IRQ1 is "also" active.
The kernel can't distinguish from a real keyboard keypress and wakes the
system.
Add the model into the list of quirks to disable keyboard wakeup source.
This is intentionally only matching the production BIOS version in hopes
that a newer EC firmware included in a newer BIOS can avoid this behavior.
platform/x86/amd/pmc: Move keyboard wakeup disablement detection to pmc-quirks
Other platforms may need to disable keyboard wakeup besides Cezanne,
so move the detection into amd_pmc_quirks_init() where it may be applied
to multiple platforms.
platform/x86/amd/pmc: Only run IRQ1 firmware version check on Cezanne
amd_pmc_wa_czn_irq1() only runs on Cezanne platforms currently but
may be extended to other platforms in the future. Rename the function
and only check platform firmware version when it's called for a Cezanne
based platform.
Rajvi Jingar [Sat, 16 Dec 2023 01:16:50 +0000 (17:16 -0800)]
platform/x86/intel/pmc: Fix hang in pmc_core_send_ltr_ignore()
For input value 0, PMC stays unassigned which causes crash while trying
to access PMC for register read/write. Include LTR index 0 in pmc_index
and ltr_index calculation.
Vishnu Sankar [Thu, 14 Dec 2023 13:47:02 +0000 (22:47 +0900)]
platform/x86: thinkpad_acpi: fix for incorrect fan reporting on some ThinkPad systems
Some ThinkPad systems ECFW use non-standard addresses for fan control
and reporting. This patch adds support for such ECFW so that it can report
the correct fan values.
Tested on Thinkpads L13 Yoga Gen 2 and X13 Yoga Gen 2.
The func_addr used to be NULL for indirect trampolines used by struct_ops.
Now func_addr is a valid function pointer.
Hence use BPF_TRAMP_F_INDIRECT flag to detect such condition.
Heiko Carstens [Fri, 8 Dec 2023 14:03:15 +0000 (15:03 +0100)]
s390/vx: fix save/restore of fpu kernel context
The KERNEL_FPR mask only contains a flag for the first eight vector
registers. However floating point registers overlay parts of the first
sixteen vector registers.
This could lead to vector register corruption if a kernel fpu context uses
any of the vector registers 8 to 15 and is interrupted or calls a
KERNEL_FPR context. If that context uses also vector registers 8 to 15,
their contents will be corrupted on return.
Luckily this is currently not a real bug, since the kernel has only one
KERNEL_FPR user with s390_adjust_jiffies() and it is only using floating
point registers 0 to 2.
Fix this by using the correct bits for KERNEL_FPR.
Fixes: 7f79695cc1b6 ("s390/fpu: improve kernel_fpu_[begin|end]") Signed-off-by: Heiko Carstens <[email protected]> Reviewed-by: Hendrik Brueckner <[email protected]> Signed-off-by: Alexander Gordeev <[email protected]>
Introduce the rcu_replace_pointer_rtnl helper to lockdep check rtnl lock
rcu replacements, alongside the already existing helpers.
Patch 2 uses the new helper in the rtnl_unregister_* functions.
Originally this change was part of the P4TC series, as it's a recurrent
pattern there, but since it has a use case in mainline we are pushing it
separately.
====================
Jamal Hadi Salim [Fri, 15 Dec 2023 17:57:10 +0000 (14:57 -0300)]
net: rtnl: introduce rcu_replace_pointer_rtnl
Introduce the rcu_replace_pointer_rtnl helper to lockdep check rtnl lock
rcu replacements, alongside the already existing helpers.
This is a quality of life helper so instead of using:
rcu_replace_pointer(rp, p, lockdep_rtnl_is_held())
.. or the open coded..
rtnl_dereference() / rcu_assign_pointer()
.. or the lazy check version ..
rcu_replace_pointer(rp, p, 1)
Use:
rcu_replace_pointer_rtnl(rp, p)
Paulo Alcantara [Wed, 13 Dec 2023 15:25:57 +0000 (12:25 -0300)]
smb: client: fix OOB in SMB2_query_info_init()
A small CIFS buffer (448 bytes) isn't big enough to hold
SMB2_QUERY_INFO request along with user's input data from
CIFS_QUERY_INFO ioctl. That is, if the user passed an input buffer >
344 bytes, the client will memcpy() off the end of @req->Buffer in
SMB2_query_info_init() thus causing the following KASAN splat:
BUG: KASAN: slab-out-of-bounds in SMB2_query_info_init+0x242/0x250 [cifs]
Write of size 1023 at addr ffff88801308c5a8 by task a.out/1240
Fix this by increasing size of SMB2_QUERY_INFO request buffers and
validating input length to prevent other callers from overflowing @req
in SMB2_query_info_init() as well.
Paulo Alcantara [Wed, 13 Dec 2023 15:25:56 +0000 (12:25 -0300)]
smb: client: fix OOB in cifsd when receiving compounded resps
Validate next header's offset in ->next_header() so that it isn't
smaller than MID_HEADER_SIZE(server) and then standard_receive3() or
->receive() ends up writing off the end of the buffer because
'pdu_length - MID_HEADER_SIZE(server)' wraps up to a huge length:
BUG: KASAN: slab-out-of-bounds in _copy_to_iter+0x4fc/0x840
Write of size 701 at addr ffff88800caf407f by task cifsd/1090
David S. Miller [Sun, 17 Dec 2023 20:54:22 +0000 (20:54 +0000)]
Merge branch 'mptcp-misc-fixes'
Matthieu Baerts says:
====================
mptcp: misc. fixes for v6.7
Here are a few fixes related to MPTCP:
Patch 1 avoids skipping some subtests of the MPTCP Join selftest by
mistake when using older versions of GCC. This fixes a patch introduced
in v6.4, backported up to v6.1.
Patch 2 fixes an inconsistent state when using MPTCP + FastOpen. A fix
for v6.2.
Patch 3 adds a description for MPTCP Kunit test modules to avoid a
warning.
Patch 4 adds an entry to the mailmap file for Geliang's email addresses.
====================
and transitioning such subflow to FIN_WAIT1 status before the syn-ack
packet is processed. The MPTCP code does not react to such state change,
leaving the connection in not-fallback status and the subflow handshake
uncompleted, triggering the following splat:
To address the issue, catch the racing subflow state change and
use it to cause the MPTCP fallback. Such fallback is also used to
cause the first subflow state propagation to the msk socket via
mptcp_set_connected(). After this change, the first subflow can
additionally propagate the TCP_FIN_WAIT1 state, so rename the
helper accordingly.
Finally, if the state propagation is delayed to the msk release
callback, the first subflow can change to a different state in between.
Cache the relevant target state in a new msk-level field and use
such value to update the msk state at release time.
MPC backups tests will skip unexpected sometimes (For example, when
compiling kernel with an older version of gcc, such as gcc-8), since
static functions like mptcp_subflow_send_ack also be listed in
/proc/kallsyms, with a 't' in front of it, not 'T' ('T' is for a global
function):
In this case, mptcp_lib_kallsyms_doesnt_have "mptcp_subflow_send_ack$"
will be false, MPC backups tests will skip. This is not what we expected.
The correct logic here should be: if mptcp_subflow_send_ack is not a
global function in /proc/kallsyms, do these MPC backups tests. So a 'T'
must be added in front of mptcp_subflow_send_ack.
This small series is required for the upcoming qca807x PHY that
will make use of PHY package mmd API and the new implementation
with read/write based on base addr.
The MMD PHY package patch currently has no use but it will be
used in the upcoming patch and it does complete what a PHY package
may require in addition to basic read/write to setup global PHY address.
(Changelog for all the revision is present in the single patch)
====================
net: phy: add support for PHY package MMD read/write
Some PHY in PHY package may require to read/write MMD regs to correctly
configure the PHY package.
Add support for these additional required function in both lock and no
lock variant.
It's assumed that the entire PHY package is either C22 or C45. We use
C22 or C45 way of writing/reading to mmd regs based on the passed phydev
whether it's C22 or C45.
net: phy: restructure __phy_write/read_mmd to helper and phydev user
Restructure phy_write_mmd and phy_read_mmd to implement generic helper
for direct mdiobus access for mmd and use these helper for phydev user.
This is needed in preparation of PHY package API that requires generic
access to the mdiobus and are deatched from phydev struct but instead
access them based on PHY package base_addr and offsets.
net: phy: extend PHY package API to support multiple global address
Current API for PHY package are limited to single address to configure
global settings for the PHY package.
It was found that some PHY package (for example the qca807x, a PHY
package that is shipped with a bundle of 5 PHY) requires multiple PHY
address to configure global settings. An example scenario is a PHY that
have a dedicated PHY for PSGMII/serdes calibrarion and have a specific
PHY in the package where the global PHY mode is set and affects every
other PHY in the package.
Change the API in the following way:
- Change phy_package_join() to take the base addr of the PHY package
instead of the global PHY addr.
- Make __/phy_package_write/read() require an additional arg that
select what global PHY address to use by passing the offset from the
base addr passed on phy_package_join().
Each user of this API is updated to follow this new implementation
following a pattern where an enum is defined to declare the offset of the
addr.
We also drop the check if shared is defined as any user of the
phy_package_read/write is expected to use phy_package_join first. Misuse
of this will correctly trigger a kernel panic for NULL pointer
exception.
Suman Ghosh [Fri, 15 Dec 2023 12:01:49 +0000 (17:31 +0530)]
octeontx2-af: Add new devlink param to configure maximum usable NIX block LFs
On some silicon variants the number of available CAM entries are
less. Reserving one entry for each NIX-LF for default DMAC based pkt
forwarding rules will reduce the number of available CAM entries
further. Hence add configurability via devlink to set maximum number of
NIX-LFs needed which inturn frees up some CAM entries.
Linus Torvalds [Sun, 17 Dec 2023 17:27:36 +0000 (09:27 -0800)]
Merge tag 'for-6.7-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fix from David Sterba:
"One more fix that verifies that the snapshot source is a root, same
check is also done in user space but should be done by the ioctl as
well"
* tag 'for-6.7-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: do not allow non subvolume root targets for snapshot
Linus Torvalds [Sun, 17 Dec 2023 17:24:06 +0000 (09:24 -0800)]
Merge tag 'soundwire-6.7-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/soundwire
Pull soundwire fixes from Vinod Koul:
- Null pointer dereference for mult link in core
- AC timing fix in intel driver
* tag 'soundwire-6.7-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/soundwire:
soundwire: intel_ace2x: fix AC timing setting for ACE2.x
soundwire: stream: fix NULL pointer dereference for multi_link
Linus Torvalds [Sun, 17 Dec 2023 17:19:27 +0000 (09:19 -0800)]
Merge tag 'phy-fixes-6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy
Pull phy fixes from Vinod Koul:
- register offset fix for TI driver
- mediatek driver minimal supported frequency fix
- negative error code in probe fix for sunplus driver
* tag 'phy-fixes-6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy:
phy: sunplus: return negative error code in sp_usb_phy_probe
phy: mediatek: mipi: mt8183: fix minimal supported frequency
phy: ti: gmii-sel: Fix register offset when parent is not a syscon node
Linus Torvalds [Sun, 17 Dec 2023 17:11:32 +0000 (09:11 -0800)]
Merge tag 'dmaengine-fix-6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine
Pull dmaengine fixes from Vinod Koul:
- SPI PDMA data fix for TI k3-psil drivers
- suspend fix, pointer check, logic for arbitration fix and channel
leak fix in fsl-edma driver
- couple of fixes in idxd driver for GRPCFG descriptions and int_handle
field handling
- single fix for stm32 driver for bitfield overflow
* tag 'dmaengine-fix-6.7' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine:
dmaengine: fsl-edma: fix DMA channel leak in eDMAv4
dmaengine: fsl-edma: fix wrong pointer check in fsl_edma3_attach_pd()
dmaengine: idxd: Fix incorrect descriptions for GRPCFG register
dmaengine: idxd: Protect int_handle field in hw descriptor
dmaengine: stm32-dma: avoid bitfield overflow assertion
dmaengine: fsl-edma: Add judgment on enabling round robin arbitration
dmaengine: fsl-edma: Do not suspend and resume the masked dma channel when the system is sleeping
dmaengine: ti: k3-psil-am62a: Fix SPI PDMA data
dmaengine: ti: k3-psil-am62: Fix SPI PDMA data
Linus Torvalds [Sun, 17 Dec 2023 17:07:34 +0000 (09:07 -0800)]
Merge tag 'cxl-fixes-6.7-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl
Pull CXL (Compute Express Link) fixes from Dan Williams:
"A collection of CXL fixes.
The touch outside of drivers/cxl/ is for a helper that allocates
physical address space. Device hotplug tests showed that the driver
failed to utilize (skipped over) valid capacity when allocating a new
memory region. Outside of that, new tests uncovered a small crop of
lockdep reports.
There is also some miscellaneous error path and leak fixups that are
not urgent, but useful to cleanup now.
- Fix put_device() like for the new CXL PMU driver
- Fix wrong pointer freed on error path
- Fixup several lockdep reports (missing lock hold) from new
assertion in cxl_num_decoders_committed() and new tests"
* tag 'cxl-fixes-6.7-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/cxl/cxl:
cxl/pmu: Ensure put_device on pmu devices
cxl/cdat: Free correct buffer on checksum error
cxl/hdm: Fix dpa translation locking
kernel/resource: Increment by align value in get_free_mem_region()
cxl: Add cxl_num_decoders_committed() usage to cxl_test
cxl/memdev: Hold region_rwsem during inject and clear poison ops
cxl/core: Always hold region_rwsem while reading poison lists
cxl/hdm: Fix a benign lockdep splat
Linus Torvalds [Sun, 17 Dec 2023 17:02:20 +0000 (09:02 -0800)]
Merge tag 'edac_urgent_for_v6.7_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras
Pull EDAC fix from Borislav Petkov:
- A single fix for the EDAC Versal driver to read out register fields
properly
* tag 'edac_urgent_for_v6.7_rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/ras/ras:
EDAC/versal: Read num_csrows and num_chans using the correct bitfield macro
Linus Torvalds [Sun, 17 Dec 2023 16:50:00 +0000 (08:50 -0800)]
Merge tag 'powerpc-6.7-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux
Pull powerpc fixes from Michael Ellerman:
- Fix a bug where heavy VAS (accelerator) usage could race with
partition migration and prevent the migration from completing.
- Update MAINTAINERS to add Aneesh & Naveen.
Thanks to Haren Myneni.
* tag 'powerpc-6.7-5' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
MAINTAINERS: powerpc: Add Aneesh & Naveen
powerpc/pseries/vas: Migration suspend waits for no in-progress open windows
Amir Goldstein [Sun, 17 Dec 2023 09:08:52 +0000 (11:08 +0200)]
ovl: fix dentry reference leak after changes to underlying layers
syzbot excercised the forbidden practice of moving the workdir under
lowerdir while overlayfs is mounted and tripped a dentry reference leak.
Fixes: c63e56a4a652 ("ovl: do not open/llseek lower file with upper sb_writers held") Reported-and-tested-by: [email protected] Signed-off-by: Amir Goldstein <[email protected]>
wifi: mt76: mt7996: Use DECLARE_FLEX_ARRAY() and fix -Warray-bounds warnings
Transform zero-length arrays `rate`, `adm_stat` and `msdu_cnt` into
proper flexible-array members in anonymous union in `struct
mt7996_mcu_all_sta_info_event` via the DECLARE_FLEX_ARRAY()
helper; and fix multiple -Warray-bounds warnings:
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:544:61: warning: array subscript <unknown> is outside array bounds of 'struct <anonymous>[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:551:58: warning: array subscript <unknown> is outside array bounds of 'struct <anonymous>[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:553:58: warning: array subscript <unknown> is outside array bounds of 'struct <anonymous>[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:530:61: warning: array subscript <unknown> is outside array bounds of 'struct <anonymous>[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:538:66: warning: array subscript <unknown> is outside array bounds of 'struct <anonymous>[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:540:66: warning: array subscript <unknown> is outside array bounds of 'struct <anonymous>[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:520:57: warning: array subscript <unknown> is outside array bounds of 'struct all_sta_trx_rate[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:526:76: warning: array subscript <unknown> is outside array bounds of 'struct all_sta_trx_rate[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:526:76: warning: array subscript <unknown> is outside array bounds of 'struct all_sta_trx_rate[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:526:76: warning: array subscript <unknown> is outside array bounds of 'struct all_sta_trx_rate[0]' [-Warray-bounds=]
drivers/net/wireless/mediatek/mt76/mt7996/mcu.c:526:76: warning: array subscript <unknown> is outside array bounds of 'struct all_sta_trx_rate[0]' [-Warray-bounds=]
This results in no differences in binary output, helps with the ongoing
efforts to globally enable -Warray-bounds.
With page pool in use, this combination can be quite common(ex.
NetworkMananger may lead to the additional packet_type being registered,
thus the cloning). In scenarios with a higher number of small packets, it
can significantly affect the success rate of coalescing.
This patchset aims to optimize this scenario and enable coalescing of this
particular combination. That also involves supporting multiple users
referencing the same fragment of a pp page to accomondate the need to
increment the "from" SKB page's pp page reference count.
Changes from v10:
- re-number patches to 1/3, 2/3, 3/3
Changes from v9:
- patch 1 was already applied
- imporve description for patch 2
- make sure skb_pp_frag_ref only work for pp aware skbs
====================
Liang Chen [Fri, 15 Dec 2023 03:30:11 +0000 (11:30 +0800)]
skbuff: Optimization of SKB coalescing for page pool
In order to address the issues encountered with commit 1effe8ca4e34
("skbuff: fix coalescing for page_pool fragment recycling"), the
combination of the following condition was excluded from skb coalescing:
However, with page pool environments, the aforementioned combination can
be quite common(ex. NetworkMananger may lead to the additional
packet_type being registered, thus the cloning). In scenarios with a
higher number of small packets, it can significantly affect the success
rate of coalescing. For example, considering packets of 256 bytes size,
our comparison of coalescing success rate is as follows:
Without page pool: 70%
With page pool: 13%
Consequently, this has an impact on performance:
Without page pool: 2.57 Gbits/sec
With page pool: 2.26 Gbits/sec
Therefore, it seems worthwhile to optimize this scenario and enable
coalescing of this particular combination. To achieve this, we need to
ensure the correct increment of the "from" SKB page's page pool
reference count (pp_ref_count).
Following this optimization, the success rate of coalescing measured in
our environment has improved as follows:
With page pool: 60%
This success rate is approaching the rate achieved without using page
pool, and the performance has also been improved:
With page pool: 2.52 Gbits/sec
Below is the performance comparison for small packets before and after
this optimization. We observe no impact to packets larger than 4K.
Liang Chen [Fri, 15 Dec 2023 03:30:09 +0000 (11:30 +0800)]
page_pool: halve BIAS_MAX for multiple user references of a fragment
Up to now, we were only subtracting from the number of used page fragments
to figure out when a page could be freed or recycled. A following patch
introduces support for multiple users referencing the same fragment. So
reduce the initial page fragments value to half to avoid overflowing.
An essential part of any big kernel submissions is selftests.
At the beginning of TCP-AO project, I made patches to fcnal-test.sh
and nettest.c to have the benefits of easy refactoring, early noticing
breakages, putting a moat around the code, documenting
and designing uAPI.
While tests based on fcnal-test.sh/nettest.c provided initial testing*
and were very easy to add, the pile of TCP-AO quickly grew out of
one-binary + shell-script testing.
The design of the TCP-AO testing is a bit different than one-big
selftest binary as I did previously in net/ipsec.c. I found it
beneficial to avoid implementing a tests runner/scheduler and delegate
it to the user or Makefile. The approach is very influenced
by CRIU/ZDTM testing[1]: it provides a static library with helper
functions and selftest binaries that create specific scenarios.
I also tried to utilize kselftest.h.
test_init() function does all needed preparations. To not leave
any traces after a selftest exists, it creates a network namespace
and if the test wants to establish a TCP connection, a child netns.
The parent and child netns have veth pair with proper ip addresses
and routes set up. Both peers, the client and server are different
pthreads. The treading model was chosen over forking mostly by easiness
of cleanup on a failure: no need to search for children, handle SIGCHLD,
make sure not to wait for a dead peer to perform anything, etc.
Any thread that does exit() naturally kills the tests, sweet!
The selftests are compiled currently in two variants: ipv4 and ipv6.
Ipv4-mapped-ipv6 addresses might be a third variant to add, but it's not
there in this version. As pretty much all tests are shared between two
address families, most of the code can be shared, too. To differ in code
what kind of test is running, Makefile supplies -DIPV6_TEST to compiler
and ifdeffery in tests can do things that have to be different between
address families. This is similar to TARGETS_C_BOTHBITS in x86 selftests
and also to tests code sharing in CRIU/ZDTM.
The total number of tests is 832.
From them rst_ipv{4,6} has currently one flaky subtest, that may fail:
> not ok 9 client connection was not reset: 0
I'll investigate what happens there. Also, unsigned-md5_ipv{4,6}
are flaky because of netns counter checks: it doesn't expect that
there may be retransmitted TCP segments from a previous sub-selftest.
That will be fixed. Besides, key-management_ipv{4,6} has 3 sub-tests
passing with XFAIL:
> ok 15 # XFAIL listen() after current/rnext keys set: the socket has current/rnext keys: 100:200
> ok 16 # XFAIL listen socket, delete current key from before listen(): failed to delete the key 100:100 -16
> ok 17 # XFAIL listen socket, delete rnext key from before listen(): failed to delete the key 200:200 -16
...
> # Totals: pass:117 fail:0 xfail:3 xpass:0 skip:0 error:0
Those need some more kernel work to pass instead of xfail.
The overview of selftests (see the diffstat at the bottom):
├── lib
│ ├── aolib.h
│ │ The header for all selftests to include.
│ ├── kconfig.c
│ │ Kernel kconfig detector to SKIP tests that depend on something.
│ ├── netlink.c
│ │ Netlink helper to add/modify/delete VETH/IPs/routes/VRFs
│ │ I considered just using libmnl, but this is around 400 lines
│ │ and avoids selftests dependency on out-of-tree sources/packets.
│ ├── proc.c
│ │ SNMP/netstat procfs parser and the counters comparator.
│ ├── repair.c
│ │ Heavily influenced by libsoccr and reduced to minimum TCP
│ │ socket checkpoint/repair. Shouldn't be used out of selftests,
│ │ though.
│ ├── setup.c
│ │ All the needed netns/veth/ips/etc preparations for test init.
│ ├── sock.c
│ │ Socket helpers: {s,g}etsockopt()s/connect()/listen()/etc.
│ └── utils.c
│ Random stuff (a pun intended).
├── bench-lookups.c
│ The only benchmark in selftests currently: checks how well TCP-AO
│ setsockopt()s perform, depending on the amount of keys on a socket.
├── connect.c
│ Trivial sample, can be used as a boilerplate to write a new test.
├── connect-deny.c
│ More-or-less what could be expected for TCP-AO in fcnal-test.sh
├── icmps-accept.c -> icmps-discard.c
├── icmps-discard.c
│ Verifies RFC5925 (7.8) by checking that TCP-AO connection can be
│ broken if ICMPs are accepted and survives when ::accept_icmps = 0
├── key-management.c
│ Key manipulations, rotations between randomized hashing algorithms
│ and counter checks for those scenarios.
├── restore.c
│ TCP_AO_REPAIR: verifies that a socket can be re-created without
│ TCP-AO connection being interrupted.
├── rst.c
│ As RST segments are signed on a separate code-path in kernel,
│ verifies passive/active TCP send_reset().
├── self-connect.c
│ Verifies that TCP self-connect and also simultaneous open work.
├── seq-ext.c
│ Utilizes TCP_AO_REPAIR to check that on SEQ roll-over SNE
│ increment is performed and segments with different SNEs fail to
│ pass verification.
├── setsockopt-closed.c
│ Checks that {s,g}etsockopt()s are extendable syscalls and common
│ error-paths for them.
└── unsigned-md5.c
Checks listen() socket for (non-)matching peers with: AO/MD5/none
keys. As well as their interaction with VRFs and AO_REQUIRED flag.
There are certainly more test scenarios that can be added, but even so,
I'm pretty happy that this much of TCP-AO functionality and uAPIs got
covered. These selftests were iteratively developed by me during TCP-AO
kernel upstreaming and the resulting kernel patches would have been
worse without having these tests. They provided the user-side
perspective but also allowed safer refactoring with less possibility
of introducing a regression. Now it's time to use them to dig
a moat around the TCP-AO code!
There are also people from other network companies that work on TCP-AO
(+testing), so sharing these selftests will allow them to contribute
and may benefit from their efforts.
====================
Dmitry Safonov [Fri, 15 Dec 2023 02:36:26 +0000 (02:36 +0000)]
selftests/net: Add TCP-AO key-management test
Check multiple keys on a socket:
- rotation on closed socket
- current/rnext operations shouldn't be possible on listen sockets
- current/rnext key set should be the one, that's used on connect()
- key rotations with pseudo-random generated keys
- copying matching keys on connect() and on accept()
At this moment there are 3 tests that are "expected" to fail: a kernel
fix is needed to improve the situation, they are marked XFAIL.
Sample output:
> # ./key-management_ipv4
> 1..120
> # 1601[lib/setup.c:239] rand seed 1700526653
> TAP version 13
> ok 1 closed socket, delete a key: the key was deleted
> ok 2 closed socket, delete all keys: the key was deleted
> ok 3 closed socket, delete current key: key deletion was prevented
> ok 4 closed socket, delete rnext key: key deletion was prevented
> ok 5 closed socket, delete a key + set current/rnext: the key was deleted
> ok 6 closed socket, force-delete current key: the key was deleted
> ok 7 closed socket, force-delete rnext key: the key was deleted
> ok 8 closed socket, delete current+rnext key: key deletion was prevented
> ok 9 closed socket, add + change current key
> ok 10 closed socket, add + change rnext key
> ok 11 listen socket, delete a key: the key was deleted
> ok 12 listen socket, delete all keys: the key was deleted
> ok 13 listen socket, setting current key not allowed
> ok 14 listen socket, setting rnext key not allowed
> ok 15 # XFAIL listen() after current/rnext keys set: the socket has current/rnext keys: 100:200
> ok 16 # XFAIL listen socket, delete current key from before listen(): failed to delete the key 100:100 -16
> ok 17 # XFAIL listen socket, delete rnext key from before listen(): failed to delete the key 200:200 -16
> ok 18 listen socket, getsockopt(TCP_AO_REPAIR) is restricted
> ok 19 listen socket, setsockopt(TCP_AO_REPAIR) is restricted
> ok 20 listen socket, delete a key + set current/rnext: key deletion was prevented
> ok 21 listen socket, force-delete current key: key deletion was prevented
> ok 22 listen socket, force-delete rnext key: key deletion was prevented
> ok 23 listen socket, delete a key: the key was deleted
> ok 24 listen socket, add + change current key
> ok 25 listen socket, add + change rnext key
> ok 26 server: Check current/rnext keys unset before connect(): The socket keys are consistent with the expectations
> ok 27 client: Check current/rnext keys unset before connect(): current key 19 as expected
> ok 28 client: Check current/rnext keys unset before connect(): rnext key 146 as expected
> ok 29 server: Check current/rnext keys unset before connect(): server alive
> ok 30 server: Check current/rnext keys unset before connect(): passed counters checks
> ok 31 client: Check current/rnext keys unset before connect(): The socket keys are consistent with the expectations
> ok 32 server: Check current/rnext keys unset before connect(): The socket keys are consistent with the expectations
> ok 33 server: Check current/rnext keys unset before connect(): passed counters checks
> ok 34 client: Check current/rnext keys unset before connect(): passed counters checks
> ok 35 server: Check current/rnext keys set before connect(): The socket keys are consistent with the expectations
> ok 36 server: Check current/rnext keys set before connect(): server alive
> ok 37 server: Check current/rnext keys set before connect(): passed counters checks
> ok 38 client: Check current/rnext keys set before connect(): current key 10 as expected
> ok 39 client: Check current/rnext keys set before connect(): rnext key 137 as expected
> ok 40 server: Check current/rnext keys set before connect(): The socket keys are consistent with the expectations
> ok 41 client: Check current/rnext keys set before connect(): The socket keys are consistent with the expectations
> ok 42 client: Check current/rnext keys set before connect(): passed counters checks
> ok 43 server: Check current/rnext keys set before connect(): passed counters checks
> ok 44 server: Check current != rnext keys set before connect(): The socket keys are consistent with the expectations
> ok 45 server: Check current != rnext keys set before connect(): server alive
> ok 46 server: Check current != rnext keys set before connect(): passed counters checks
> ok 47 client: Check current != rnext keys set before connect(): current key 10 as expected
> ok 48 client: Check current != rnext keys set before connect(): rnext key 132 as expected
> ok 49 server: Check current != rnext keys set before connect(): The socket keys are consistent with the expectations
> ok 50 client: Check current != rnext keys set before connect(): The socket keys are consistent with the expectations
> ok 51 client: Check current != rnext keys set before connect(): passed counters checks
> ok 52 server: Check current != rnext keys set before connect(): passed counters checks
> ok 53 server: Check current flapping back on peer's RnextKey request: The socket keys are consistent with the expectations
> ok 54 server: Check current flapping back on peer's RnextKey request: server alive
> ok 55 server: Check current flapping back on peer's RnextKey request: passed counters checks
> ok 56 client: Check current flapping back on peer's RnextKey request: current key 10 as expected
> ok 57 client: Check current flapping back on peer's RnextKey request: rnext key 132 as expected
> ok 58 server: Check current flapping back on peer's RnextKey request: The socket keys are consistent with the expectations
> ok 59 client: Check current flapping back on peer's RnextKey request: The socket keys are consistent with the expectations
> ok 60 server: Check current flapping back on peer's RnextKey request: passed counters checks
> ok 61 client: Check current flapping back on peer's RnextKey request: passed counters checks
> ok 62 server: Rotate over all different keys: The socket keys are consistent with the expectations
> ok 63 server: Rotate over all different keys: server alive
> ok 64 server: Rotate over all different keys: passed counters checks
> ok 65 server: Rotate over all different keys: current key 128 as expected
> ok 66 client: Rotate over all different keys: rnext key 128 as expected
> ok 67 server: Rotate over all different keys: current key 129 as expected
> ok 68 client: Rotate over all different keys: rnext key 129 as expected
> ok 69 server: Rotate over all different keys: current key 130 as expected
> ok 70 client: Rotate over all different keys: rnext key 130 as expected
> ok 71 server: Rotate over all different keys: current key 131 as expected
> ok 72 client: Rotate over all different keys: rnext key 131 as expected
> ok 73 server: Rotate over all different keys: current key 132 as expected
> ok 74 client: Rotate over all different keys: rnext key 132 as expected
> ok 75 server: Rotate over all different keys: current key 133 as expected
> ok 76 client: Rotate over all different keys: rnext key 133 as expected
> ok 77 server: Rotate over all different keys: current key 134 as expected
> ok 78 client: Rotate over all different keys: rnext key 134 as expected
> ok 79 server: Rotate over all different keys: current key 135 as expected
> ok 80 client: Rotate over all different keys: rnext key 135 as expected
> ok 81 server: Rotate over all different keys: current key 136 as expected
> ok 82 client: Rotate over all different keys: rnext key 136 as expected
> ok 83 server: Rotate over all different keys: current key 137 as expected
> ok 84 client: Rotate over all different keys: rnext key 137 as expected
> ok 85 server: Rotate over all different keys: current key 138 as expected
> ok 86 client: Rotate over all different keys: rnext key 138 as expected
> ok 87 server: Rotate over all different keys: current key 139 as expected
> ok 88 client: Rotate over all different keys: rnext key 139 as expected
> ok 89 server: Rotate over all different keys: current key 140 as expected
> ok 90 client: Rotate over all different keys: rnext key 140 as expected
> ok 91 server: Rotate over all different keys: current key 141 as expected
> ok 92 client: Rotate over all different keys: rnext key 141 as expected
> ok 93 server: Rotate over all different keys: current key 142 as expected
> ok 94 client: Rotate over all different keys: rnext key 142 as expected
> ok 95 server: Rotate over all different keys: current key 143 as expected
> ok 96 client: Rotate over all different keys: rnext key 143 as expected
> ok 97 server: Rotate over all different keys: current key 144 as expected
> ok 98 client: Rotate over all different keys: rnext key 144 as expected
> ok 99 server: Rotate over all different keys: current key 145 as expected
> ok 100 client: Rotate over all different keys: rnext key 145 as expected
> ok 101 server: Rotate over all different keys: current key 146 as expected
> ok 102 client: Rotate over all different keys: rnext key 146 as expected
> ok 103 server: Rotate over all different keys: current key 127 as expected
> ok 104 client: Rotate over all different keys: rnext key 127 as expected
> ok 105 client: Rotate over all different keys: current key 0 as expected
> ok 106 client: Rotate over all different keys: rnext key 127 as expected
> ok 107 server: Rotate over all different keys: The socket keys are consistent with the expectations
> ok 108 client: Rotate over all different keys: The socket keys are consistent with the expectations
> ok 109 client: Rotate over all different keys: passed counters checks
> ok 110 server: Rotate over all different keys: passed counters checks
> ok 111 server: Check accept() => established key matching: The socket keys are consistent with the expectations
> ok 112 Can't add a key with non-matching ip-address for established sk
> ok 113 Can't add a key with non-matching VRF for established sk
> ok 114 server: Check accept() => established key matching: server alive
> ok 115 server: Check accept() => established key matching: passed counters checks
> ok 116 client: Check connect() => established key matching: current key 0 as expected
> ok 117 client: Check connect() => established key matching: rnext key 128 as expected
> ok 118 client: Check connect() => established key matching: The socket keys are consistent with the expectations
> ok 119 server: Check accept() => established key matching: The socket keys are consistent with the expectations
> ok 120 server: Check accept() => established key matching: passed counters checks
> # Totals: pass:120 fail:0 xfail:0 xpass:0 skip:0 error:0
Dmitry Safonov [Fri, 15 Dec 2023 02:36:25 +0000 (02:36 +0000)]
selftests/net: Add TCP-AO selfconnect/simultaneous connect test
Check that a rare functionality of TCP named self-connect works with
TCP-AO. This "under the cover" also checks TCP simultaneous connect
(TCP_SYN_RECV socket state), which would be harder to check other ways.
In order to verify that it's indeed TCP simultaneous connect, check
the counters TCPChallengeACK and TCPSYNChallenge.
Sample of the output:
> # ./self-connect_ipv6
> 1..4
> # 1738[lib/setup.c:254] rand seed 1696451931
> TAP version 13
> ok 1 self-connect(same keyids): connect TCPAOGood 0 => 24
> ok 2 self-connect(different keyids): connect TCPAOGood 26 => 50
> ok 3 self-connect(restore): connect TCPAOGood 52 => 97
> ok 4 self-connect(restore, different keyids): connect TCPAOGood 99 => 144
> # Totals: pass:4 fail:0 xfail:0 xpass:0 skip:0 error:0
Dmitry Safonov [Fri, 15 Dec 2023 02:36:24 +0000 (02:36 +0000)]
selftests/net: Add TCP-AO RST test
Check that both active and passive reset works and correctly sign
segments with TCP-AO or don't send RSTs if not possible to sign.
A listening socket with backlog = 0 gets one connection in accept
queue, another in syn queue. Once the server/listener socket is
forcibly closed, client sockets aren't connected to anything.
In regular situation they would receive RST on any segment, but
with TCP-AO as there's no listener, no AO-key and unknown ISNs,
no RST should be sent.
And "passive" reset, where RST is sent on reply for some segment
(tcp_v{4,6}_send_reset()) - there use TCP_REPAIR to corrupt SEQ numbers,
which later results in TCP-AO signed RST, which will be verified and
client socket will get EPIPE.
No TCPAORequired/TCPAOBad segments are expected during these tests.
Sample of the output:
> # ./rst_ipv4
> 1..15
> # 1462[lib/setup.c:254] rand seed 1686611171
> TAP version 13
> ok 1 servered 1000 bytes
> ok 2 Verified established tcp connection
> ok 3 sk[0] = 7, connection was reset
> ok 4 sk[1] = 8, connection was reset
> ok 5 sk[2] = 9
> ok 6 MKT counters are good on server
> ok 7 Verified established tcp connection
> ok 8 client connection broken post-seq-adjust
> ok 9 client connection was reset
> ok 10 No segments without AO sign (server)
> ok 11 Signed AO segments (server): 0 => 30
> ok 12 No segments with bad AO sign (server)
> ok 13 No segments without AO sign (client)
> ok 14 Signed AO segments (client): 0 => 30
> ok 15 No segments with bad AO sign (client)
> # Totals: pass:15 fail:0 xfail:0 xpass:0 skip:0 error:0
Dmitry Safonov [Fri, 15 Dec 2023 02:36:22 +0000 (02:36 +0000)]
selftests/net: Add TCP_REPAIR TCP-AO tests
The test plan is:
1. check that TCP-AO connection may be restored on another socket
2. check restore with wrong send/recv ISN (checking that they are
part of MAC generation)
3. check restore with wrong SEQ number extension (checking that
high bytes of it taken into MAC generation)
Sample output expected:
> # ./restore_ipv4
> 1..20
> # 1412[lib/setup.c:254] rand seed 1686610825
> TAP version 13
> ok 1 TCP-AO migrate to another socket: server alive
> ok 2 TCP-AO migrate to another socket: post-migrate connection is alive
> ok 3 TCP-AO migrate to another socket: counter TCPAOGood increased 23 => 44
> ok 4 TCP-AO migrate to another socket: counter TCPAOGood increased 22 => 42
> ok 5 TCP-AO with wrong send ISN: server couldn't serve
> ok 6 TCP-AO with wrong send ISN: post-migrate connection is broken
> ok 7 TCP-AO with wrong send ISN: counter TCPAOBad increased 0 => 4
> ok 8 TCP-AO with wrong send ISN: counter TCPAOBad increased 0 => 3
> ok 9 TCP-AO with wrong receive ISN: server couldn't serve
> ok 10 TCP-AO with wrong receive ISN: post-migrate connection is broken
> ok 11 TCP-AO with wrong receive ISN: counter TCPAOBad increased 4 => 8
> ok 12 TCP-AO with wrong receive ISN: counter TCPAOBad increased 5 => 10
> ok 13 TCP-AO with wrong send SEQ ext number: server couldn't serve
> ok 14 TCP-AO with wrong send SEQ ext number: post-migrate connection is broken
> ok 15 TCP-AO with wrong send SEQ ext number: counter TCPAOBad increased 9 => 10
> ok 16 TCP-AO with wrong send SEQ ext number: counter TCPAOBad increased 11 => 19
> ok 17 TCP-AO with wrong receive SEQ ext number: post-migrate connection is broken
> ok 18 TCP-AO with wrong receive SEQ ext number: server couldn't serve
> ok 19 TCP-AO with wrong receive SEQ ext number: counter TCPAOBad increased 10 => 18
> ok 20 TCP-AO with wrong receive SEQ ext number: counter TCPAOBad increased 20 => 23
> # Totals: pass:20 fail:0 xfail:0 xpass:0 skip:0 error:0
projects / linux.git / log