Mitchell Horne [Thu, 13 May 2021 18:57:37 +0000 (15:57 -0300)]
libpmc: fall-back to kernel tables if pmu-events fails
On x86, the pmu_events table is the source of truth for finding
supported events. However, events not found there may still be present
in the kernel's static event tables. For example, the pmc.soft(3) events
will never be available from pmu-events.
Update pmc_allocate() to search the legacy event tables if
pmc_pmu_pmcallocate() fails to return a result. This allows both event
sources to be consulted before giving up, thus restoring pmc.soft(3) and
pmc.tsc(3) on x86.
Reviewed by: emaste
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D30216
Mitchell Horne [Thu, 13 May 2021 19:00:07 +0000 (16:00 -0300)]
libpmc: remove PMC_MDEP_TABLE logic
This logic was added for handling some of the complicated relationships
between events and x86 CPU models. Since that logic has been mostly
removed from libpmc(3) in favor of pmu-events, this no longer serves
much of a purpose. Mapping CPU types to event tables is already handled
by the switch statement in pmc_init().
Reviewed by: ray, emaste
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D30196
Andrew Turner [Thu, 1 Apr 2021 14:38:09 +0000 (14:38 +0000)]
arm64: Fix finding the pmc event ID
The lower pmc event bits were masked off to find the PMC event ID.
The doesn't work when there are more events. Switch it to use the
offser relative to the first event while also checking the ID is
in the expected range.
Reviewed by: gnn, ray
Sponsored by: Innovate UK
Differential Revision: https://reviews.freebsd.org/D29600
Mark Johnston [Thu, 15 Jul 2021 16:23:04 +0000 (12:23 -0400)]
eli: Zero pad bytes that arise when certain auth algorithms are used
When authentication is configured, GELI ensures that the amount of data
per sector is a multiple of 16 bytes. This is done in
eli_metadata_softc(). When the digest size is not a multiple of 16
bytes, this leaves some extra pad bytes at the end of every sector, and
they were not being zeroed before being written to disk. In particular,
this happens with the HMAC/SHA1, HMAC/RIPEMD160 and HMAC/SHA384 data
authentication algorithms.
This change ensures that they are zeroed before being written to disk.
Reported by: KMSAN
Reviewed by: delphij, asomers
Sponsored by: The FreeBSD Foundation
Yang Zhong [Thu, 22 Jul 2021 17:16:01 +0000 (13:16 -0400)]
mmc: Drain the intrhook in mmc_detach()
Buggy SD card drivers may attach and detach a mmc(4) driver instance in
quick succession. In this case mmc(4) must disestablish its intrhook
callback during detach. Thus, this change adds a call to
config_intrhook_drain(), which blocks or does nothing if the intrhook is
running or has already ran (the SD card was plugged in), and
disestablishes the hook if it hasn't ran yet (the SD card was not
plugged in).
PR: 254373
Reviewed by: imp, manu, markj
Sponsored by: The FreeBSD Foundation
Mark Johnston [Thu, 15 Jul 2021 16:18:17 +0000 (12:18 -0400)]
nfsclient: Avoid copying uninitialized bytes into statfs
hst will be nul-terminated but the remaining space in the buffer is left
uninitialized. Avoid copying the entire buffer to ensure that
uninitialized bytes are not leaked via statfs(2).
Reported by: KMSAN
Reviewed by: rmacklem
Sponsored by: The FreeBSD Foundation
Previously it would try to load linux.ko instead of linux64.ko
and fail. While here, don't try to match 'linuxaout'; even if
implemented, it's the same module as `linuxelf`.
Alan Somers [Thu, 20 May 2021 01:10:15 +0000 (19:10 -0600)]
fusefs: correctly set lock owner during FUSE_SETLK
During FUSE_SETLK, the owner field should uniquely identify the calling
process. The fusefs module now sets it to the process's pid.
Previously, it expected the calling process to set it directly, which
was wrong.
libfuse also apparently expects the owner field to be set during
FUSE_GETLK, though I'm not sure why.
We failed to list the new pf_syncookies.c file in sys/conf/files. This
worked for the usual configurations, where pf is a module, but not for
LINT builds.
Kristof Provost [Wed, 2 Jun 2021 16:16:03 +0000 (18:16 +0200)]
pf tests: Forwarding syncookie test
Test syncookies on a forwarding host. That is, in a setup where the
machine (or vnet) running pf is not the same as the machine (or vnet)
running the server it's protecting.
Kristof Provost [Thu, 20 May 2021 09:54:41 +0000 (11:54 +0200)]
pf: syncookie support
Import OpenBSD's syncookie support for pf. This feature help pf resist
TCP SYN floods by only creating states once the remote host completes
the TCP handshake rather than when the initial SYN packet is received.
This is accomplished by using the initial sequence numbers to encode a
cookie (hence the name) in the SYN+ACK response and verifying this on
receipt of the client ACK.
Mark Johnston [Tue, 13 Jul 2021 21:45:57 +0000 (17:45 -0400)]
gmirror: Zero the metadata block before writing
The mirror metadata fields contain string buffers and pad bytes, neither
were being zeroed before metadata was written to disk. Also, the
metadata structure is smaller than the sector size, and in one case
gmirror was failing to zero-fill the full buffer before writing.
Fix these problems by pre-zeroing the metadata structure and the sector
buffer.
Reported by: KMSAN
Sponsored by: The FreeBSD Foundation
Mark Johnston [Tue, 13 Jul 2021 21:47:27 +0000 (17:47 -0400)]
blist: Correct the node count computed in blist_create()
Commit bb4a27f927a1 added the ability to allocate a span of blocks
crossing a meta node boundary. To ensure that blst_next_leaf_alloc()
does not walk past the end of the tree, an extra all-zero meta node
needs to be present at the end of the allocation, and
blst_next_leaf_alloc() is implemented such that the presence of this
node terminates the search.
blist_create() computes the number of nodes required. It had two
problems:
1. When the size of the blist is a power of BLIST_RADIX, we would
unnecessarily allocate an extra level in the tree.
2. When the size of the blist is a multiple of BLIST_RADIX, we would
fail to allocate a terminator node. In this case,
blst_next_leaf_alloc() could scan beyond the bounds of the
allocation. This was found using KASAN.
Modify blist_create() to handle these cases correctly.
Mark Johnston [Tue, 13 Jul 2021 21:45:49 +0000 (17:45 -0400)]
fifo: Explicitly initialize generation numbers when opening
The fi_rgen and fi_wgen fields are generation numbers used when sleeping
waiting for the other end of the fifo to be opened. The fields were not
explicitly initialized after allocation, but this was harmless. To
avoid false positives from KMSAN, though, ensure that they get
initialized to zero.
Reported by: KMSAN
Sponsored by: The FreeBSD Foundation
Mark Johnston [Tue, 13 Jul 2021 21:49:39 +0000 (17:49 -0400)]
uart: Fix an out-of-bounds read in ns8250_bus_probe()
The problem is that ns8250_bus_probe() accesses a field from the
ns8250_softc, which embeds the generic UART softc, but the ns8250_softc
hasn't yet been allocated because we're still probing.
This is a regression from commit 0aefb0a63c50. This fixed a problem
where one of the upper four IER bits, which are usually reserved, needs
to be set in order to get RX interrupts before the RX FIFO is full. At
the same time, we avoid clearing those reserved bits (see commit 58957d87173, though other UART drivers I looked at do not bother with
this).
So, copy what ns8250_init() does to disable interrupts, since we don't
know what the "right" mask is at this point.
Reported by: syzbot+f256beefd0df9eb796e7@syzkaller.appspotmail.com
Reviewed by: imp
Sponsored by: The FreeBSD Foundation
When OFED was upgraded to Linux v4.9, a bunch of Linux-specific
netlink changes were dropped. Unfortunately, there was a mismerge
in this process and as a result ib_sa_cancel_query() would fail to
cancel an outstanding MAD.
This was causing rdma_destroy_id() to hang indefinitely waiting
for the MAD to complete and release the final reference.
ipoib: Fix for accessing uninitialized pointers and freed memory during attach and detach.
Call infiniband_ifdetach() early to stop ifioctl(9) calls from user-space
during device removal. Also make sure that ifioctl(9) calls are blocked from
executing until the device is fully initialized. Ideally we would delay the
infiniband_ifattach() call, but because part of the initialization is to update
the link level address, that is not possible without more significant changes.
mlx4: Map core_clock page to user space only when allowed
Currently when we map the hca_core_clock page to the user space,
there are vulnerable registers, one of which is semaphore, on
this page as well. If user read the wrong offset, it can modify the
above semaphore and hang the device.
Hence, mapping the hca_core_clock page to the user space only when
user required it specifically.
After this patch, mlx4 core_clock won't be mapped to user space by
default. Oppose to current state, where mlx4 core_clock is always mapped
to user space.
mlx4ib and mlx5ib: Set slid to zero in Ethernet completion struct
IB spec says that a lid should be ignored when link layer is Ethernet,
for example when building or parsing a CM request message (CA17-34).
However, since ib_lid_be16() and ib_lid_cpu16() validates the slid,
not only when link layer is IB, we set the slid to zero to prevent
false warnings in the kernel log.
mlx5core: Don't query the PCI config space for offline during a firmware command.
Querying the PCI config space for offline for every firmware command blocks
the PCI bus and affects performance. Especially for packet pacing and TLS
when objects are frequently created and destroyed.
ibcore: Declare ib_post_send() and ib_post_recv() arguments const
Since neither ib_post_send() nor ib_post_recv() modify the data structure
their second argument points at, declare that argument const. This change
makes it necessary to declare the 'bad_wr' argument const too and also to
modify all ULPs that call ib_post_send(), ib_post_recv() or
ib_post_srq_recv(). This patch does not change any functionality but makes
it possible for the compiler to verify whether the
ib_post_(send|recv|srq_recv) really do not modify the posted work request.
Expose ib_ucontext from a given ib_uverbs_file. Drivers that use the ioctl(9)
API may have the ib_uverbs_file and need a way to get the related ib_ucontext
from it, this is enabled by this patch.
ibcore: Clean up INIT_UDATA() and INIT_UDATA_BUF_OR_NULL() macro usage.
We get a harmless warning about the fact that we use the result of a
multiplication as a condition in INIT_UDATA_BUF_OR_NULL():
uverbs_main.c: In function 'ib_uverbs_write':
error: '*' in boolean context, suggest '&&' instead [-Werror=int-in-bool-context]
This avoids the problem by using an inline function in place of
the macro.
After changing INIT_UDATA_BUF_OR_NULL() to an inline function,
do the same change to INIT_UDATA() for consistency.
Using an inline function gives us better type safety here among other
issues with macros. I'm using u64_to_user_ptr() to convert the user
pointer to simplify the logic rather than adding lots of new type casts.
All callers to ib_modify_qp_is_ok() provides enum ib_qp_state makes the
checks of out-of-scope redundant. Let's remove them together with updating
function signature to return boolean result.
While at it remove unused "ll" parameter from ib_modify_qp_is_ok().
Before idr_remove(), ucma_get_ctx() could still find the ctx
and after rdma_destroy_id(), rdma_resolve_addr() may still
access id_priv pointer. Also, ucma_put_ctx() may use ctx after
ucma_free_ctx() too.
ucma_close() should call ucma_put_ctx() too which tests the
refcnt and waits for the last one releasing it. The similar
pattern is already used by ucma_destroy_id().
When IPoIB receives an SM LID change event, it reacts by flushing its
path record cache and rejoining multicast groups. This is the same
behavior it performs when it receives a reregistration event. This
behavior is unnecessary as an SM may have database backup or
synchronization mechanisms which permit the SM location or LID to change
without loss of multicast membership and without impact to path records.
Both opensm and the OPA FM issue reregistration events if a new SM is
started (or restarted with a new config) or an SM event occurs which
results in loss of multicast membership records by the SM (such as
opensm failover) or the SM encounters new nodes with Active ports (such
as after joining 2 fabrics by connecting switches via ISLs). Hence this
event can be depended on as the trigger for IPoIB cache and multicast
flushing.
It appears that some drivers, such as qib, and hfi1 issue the
IB_EVENT_SM_CHANGE but other drivers such as mlx4 and mlx5 do not.
Empirical testing on Mellanox EDR using ibv_asyncwatch has confirmed
that Mellanox EDR HCAs do not generate SM change events and that opensm
does generate reregistration.
An SM LID change event is generated by the mentioned drivers to reflect
that sm_lid and/or sm_sl in the local port info has changed. The intent
of this event is to permit applications and ULPs which have a local copy
of this information (or an address handle using it) to update their
information.
The intent is that the reregistration event (caused by the SM via a bit
in Set(PortInfo)) be used to inform nodes that they need to rejoin
multicast groups, resubscribe for notices and potentially update path
records.
When an SM migrates or fails over, a SM LID change event can occur. In
response IPoIB discards path records and multicast membership and loses
connectivity until these records are restored via SA requests. In very
large fabrics, it may take minutes for the SM to be ready and for the SA
responses to be supplied. This can result in undesirable and
unnecessary IPoIB connectivity impacts. It also can result in an
unnecessary storm of SA queries from all nodes in a cluster potentially
followed by yet another storm if the SM issues the reregistration
request.
The fact the Mellanox HCAs do not even generate this event, is further
evidence that on modern IB fabrics there will be no ill side effects
from the proposed changes below to reduce the reaction by 3 kernel
components to this event. So these changes should be benign for Mellanox
IB fabrics and will benefit OPA fabrics while also making ib_core and
ULP behavor "correct" as intended by the IBTA spec and kernel RDMA event
APIs.
Address these issues by removing IB_EVENT_SM_CHANGE handling from ipoib.
IPoIB does not locally store sm_lid nor sm_sl, so it does not need to do
anything on SM LID change. IPoIB makes use of other ib_core components
to issue SA requests for it and those components correctly track SM LID
and SM LID changes.
Also in ib_core multicast handling, remove the test for
IB_EVENT_SM_CHANGE. This code is moving all multicast groups to the
error state, which will trigger rejoins. This code is used by IPoIB as
well as the connection manager and other clients of multicast groups.
This kernel module centralizes group membership status and joins since a
node can only join a given group once but multiple ULPs or applications
may want to join the same group. It makes use of the sa_query.c
component in ib_core, which correctly trackes SM LID and SL. This
component does not track SM LID nor SL itself and hence need not react
to their changes.
Similarly in the ib_core cache code remove the handling for the
IB_EVENT_SM_CHANGE. In this function. The ib_cache_update function
which is ultimately called is updating local copies of the pkey table,
gid table and lmc. It does not update nor retain sm_lid nor sm_sl. As
such it does not need to be called on an SM LID change. It technically
also does not need to be called on a reregistration. The LID_CHANGE,
PKEY_CHANGE, GID_CHANGE and port state change events (PORT_ERR,
PORT_ACTICE) should be sufficient triggers.
It is worth noting that the alternative of simply having the hfi1 and
qib drivers not generate the SM LID change event was explored. While
this would duplicate what Mellanox drivers do now, it is not the correct
behavior and removes the ability for an SM to migrate without requiring
reregistration. Since both opensm and OPA SM have mechanisms to backup
or synchronize registration information, it is desirable to let them
perform SM migrations (with LID or SL changes) without requiring
reregistration when they deem it appropriate.
The problem was that the MAD PD was deallocated before the MAD CQ.
There was completion work pending for the CQ when the PD got deallocated.
When the mad completion handling reached procedure
ib_mad_post_receive_mads(), we got a use-after-free bug in the following
line of code in that procedure:
sg_list.lkey = qp_info->port_priv->pd->local_dma_lkey;
(the pd pointer in the above line is no longer valid, because the
pd has been deallocated).
We fix this by allocating the PD before the CQ in procedure
ib_mad_port_open(), and deallocating the PD after freeing the CQ
in procedure ib_mad_port_close().
Since the CQ completion work queue is flushed during ib_free_cq(),
no completions will be pending for that CQ when the PD is later
deallocated.
Note that freeing the CQ before deallocating the PD is the practice
in the ULPs.
When requested QP type is not supported for a {device, port}, return the
error right away before validating all parameters during mad agent
registration time.
ibcore: Protect against concurrent access to hardware stats.
Currently access to hardware stats buffer isn't protected, this can
result in multiple writes and reads at the same time to the same
memory location. This can lead to providing an incorrect value to
the user. Add a mutex to protect against it.
If the provider driver (such as rdma_rxe) doesn't support PMA counters,
avoid exposing its directory similar to optional hw_counters directory.
If core fails to read the PMA counter, return an error so that user can
retry later if needed.
ibcore: Fix unable to change lifespan entry for hw_counters.
This patch fixes the case where 'lifespan' entry of the hw_counters
is not writable. Currently write callback is not exposed for for
the hw_counters sysfs operation. Due to this, modifying lifespan
value results into permission denied error in below example.
ibcore: Issue DREQ when receiving REQ/REP for stale QP.
From "InfiBand Architecture Specifications Volume 1":
A QP is said to have a stale connection when only one side has
connection information. A stale connection may result if the remote CM
had dropped the connection and sent a DREQ but the DREQ was never
received by the local CM. Alternatively the remote CM may have lost
all record of past connections because its node crashed and rebooted,
while the local CM did not become aware of the remote node's reboot
and therefore did not clean up stale connections.
And:
A local CM may receive a REQ/REP for a stale connection. It shall
abort the connection issuing REJ to the REQ/REP. It shall then issue
DREQ with "DREQ:remote QPN" set to the remote QPN from the REQ/REP.
This patch solves a problem with reuse of QPN. Current codebase, that
is IPoIB, relies on a REAP-mechanism to do cleanup of the structures
in CM. A problem with this is the timeconstants governing this
mechanism; they are up to 768 seconds and the interface may look
inresponsive in that period. Issuing a DREQ (and receiving a DREP)
does the necessary cleanup and the interface comes up.
ibcore: Move debug counters to be under relevant IB device
The sysfs layout is created by CM incorrectly presented RDMA devices with
InfiniBand link layer. Layout of such devices represents device tree of
connections. By moving CM statistics to be under relevant port of IB
device, we will fix the following issues:
* Symlink name - It used device name instead of specific identifier.
* Target location - It was supposed to point to PCI-ID/infiniband_cm/
instead of PCI-ID/infiniband/
* Target name - It created extra device file under already existing
device folder, e.g. mlx5_0/mlx5_0
* Crash during boot with RDMA persistent naming patches.
ibcore: Block processing of alternate path handling in RoCE RX CM messages.
Due to the below reasons, it is better to not support alternate path receive
messages for RoCE in near term.
1. Alternate path for RoCE is not supported at rdmacm layer.
2. It is not supported in uverbs/core layer for RoCE.
3. Alternate path for IPv6 for link local address cannot resolve route
determinstically without a valid incoming interface ID whose usecase
make sense only with dual port mode.
4. init_av_from_path while processing LAP messages for IB and RoCE can
lead to adding duplicate entry of AV into the port list, leads to list
corruption.
5. rdma-core userspace a well known userspace implementation has removed
support of libucm which use ucm.ko module, which is the only module that
can trigger alternate path related messages.
6. ucm kernel module is requested to be removed from the IB core in
the following patch, https://patchwork.kernel.org/patch/10268503/ .
ibcore: Store and restore ah_attr during LAP msg processing.
During CM LAP processing, ah_attr is reinitialized on receiving
a LAP request. First likely during CM request processing.
ah_attr might get zeroed out if LAP processing fails.
Therefore, try to create a new ah_attr for the LAP message.
If the initialization fails, continue with older ah_attr.
If the initialization passes, consider the new ah_attr by
overwriting the older one.
mlx5ib: Fix XRC QP support after introducing extended atomic.
Extended atomics are supported with RC and XRC QP types, but Linux commit a60109dc9a95 added an unneeded check to to_mlx5_access_flags().
This broke XRC QPs.
The following ib_atomic_bw invocation over XRC reproduces the issue:
ib_atomic_bw -d mlx5_1 --connection=XRC --atomic_type=FETCH_AND_ADD
It is safe to remove such checks because the QP type was already checked
in ib_modify_qp_is_ok(), which was previously called from
mlx5_ib_modify_qp().
ibcore: Remove unused and erroneous msg sequence encoding.
In cm_form_tid(), a two bit message sequence number is OR'ed into bit
31-30 of the lower TID value.
After Linux commit f06d26537559 ("IB/cm: Randomize starting comm ID"), the
local_id is XOR'ed with a 32-bit random value. Hence, bit 31-30 in the
lower TID now has an arbitrarily value and it makes no sense to OR in
the message sequence number.
Adding to that, the evolution in use of IDR routines in cm_alloc_id()
has always had the possibility of returning a value with bit 30 set.
In addition, said bits are never checked.
Hence, remove the encoding and the corresponding enum.
Until today, we didn't enforce this requirement in the code, and therefore,
if we got a page size larger than 2GB, we have passed zeros in the
log_page_shift instead of the actual value and the registration failed.
This patch limits the driver to use compound pages of 2GB for mkeys.
The patch simplifies mlx5_ib_cont_pages and fixes the following
issues in the original implementation:
First issues is related to alignment of the PFNs. After the check
base + p != PFN, the alignment of the PFN wasn't checked. So the PFN
sequence 0, 1, 1, 2 would result in a page_shift of 13 even though
the 3rd PFN is not 8KB aligned.
This wasn't actually a bug because it was supported by all the
existing mlx5 compatible device, but we don't want to require
this support in all future devices.
Another issue is because the inner loop didn't advance PFN so
the test "if (base + p != pfn)" always failed for SGE with
len > (1<<page_shift).
mlx5en: Add more error checks in the transmit path.
- Upon error more completion events than requested may be generated,
particularly when using the completion event factor feature.
- Count number of event errors in the transmit path.
mlx5ib: Support RAW Ethernet when RoCE is disabled in mlx5ib(4)
On some environments, such as certain SRIOV VF configurations, RoCE is
not supported for mlx5 Ethernet ports. Currently, the driver will not
open IB device on that port.
This is problematic, since we do want user-space RAW Ethernet (RAW_PACKET
QPs) functionality to remain in place. For that end, enhance the relevant
driver flows such that we do create a device instance in that case.
mlx5en: Add missing media types for 100GBit/s, 200Gbit/s and 400Gbit/s.
Make the mlx5e_mode_table[] array one dimensional, because there is only
one entry, 10G ER/LR, which share the same protocol bit.
This patch only adds support for basic sub-type distinguishing for the
extended protocol bits. Use verbose ifconfig eeprom output to get actual
media type.
Remove write only "connector_type" variable while at it.
projects / FreeBSD / FreeBSD.git / log