| CVE |
Vendors |
Products |
Updated |
CVSS v3.1 |
| In the Linux kernel, the following vulnerability has been resolved:
zram: fix slot write race condition
Parallel concurrent writes to the same zram index result in leaked
zsmalloc handles. Schematically we can have something like this:
CPU0 CPU1
zram_slot_lock()
zs_free(handle)
zram_slot_lock()
zram_slot_lock()
zs_free(handle)
zram_slot_lock()
compress compress
handle = zs_malloc() handle = zs_malloc()
zram_slot_lock
zram_set_handle(handle)
zram_slot_lock
zram_slot_lock
zram_set_handle(handle)
zram_slot_lock
Either CPU0 or CPU1 zsmalloc handle will leak because zs_free() is done
too early. In fact, we need to reset zram entry right before we set its
new handle, all under the same slot lock scope. |
| In the Linux kernel, the following vulnerability has been resolved:
um: virtio_uml: Fix use-after-free after put_device in probe
When register_virtio_device() fails in virtio_uml_probe(),
the code sets vu_dev->registered = 1 even though
the device was not successfully registered.
This can lead to use-after-free or other issues. |
| In the Linux kernel, the following vulnerability has been resolved:
ksmbd: smbdirect: verify remaining_data_length respects max_fragmented_recv_size
This is inspired by the check for data_offset + data_length. |
| In the Linux kernel, the following vulnerability has been resolved:
qed: Don't collect too many protection override GRC elements
In the protection override dump path, the firmware can return far too
many GRC elements, resulting in attempting to write past the end of the
previously-kmalloc'ed dump buffer.
This will result in a kernel panic with reason:
BUG: unable to handle kernel paging request at ADDRESS
where "ADDRESS" is just past the end of the protection override dump
buffer. The start address of the buffer is:
p_hwfn->cdev->dbg_features[DBG_FEATURE_PROTECTION_OVERRIDE].dump_buf
and the size of the buffer is buf_size in the same data structure.
The panic can be arrived at from either the qede Ethernet driver path:
[exception RIP: qed_grc_dump_addr_range+0x108]
qed_protection_override_dump at ffffffffc02662ed [qed]
qed_dbg_protection_override_dump at ffffffffc0267792 [qed]
qed_dbg_feature at ffffffffc026aa8f [qed]
qed_dbg_all_data at ffffffffc026b211 [qed]
qed_fw_fatal_reporter_dump at ffffffffc027298a [qed]
devlink_health_do_dump at ffffffff82497f61
devlink_health_report at ffffffff8249cf29
qed_report_fatal_error at ffffffffc0272baf [qed]
qede_sp_task at ffffffffc045ed32 [qede]
process_one_work at ffffffff81d19783
or the qedf storage driver path:
[exception RIP: qed_grc_dump_addr_range+0x108]
qed_protection_override_dump at ffffffffc068b2ed [qed]
qed_dbg_protection_override_dump at ffffffffc068c792 [qed]
qed_dbg_feature at ffffffffc068fa8f [qed]
qed_dbg_all_data at ffffffffc0690211 [qed]
qed_fw_fatal_reporter_dump at ffffffffc069798a [qed]
devlink_health_do_dump at ffffffff8aa95e51
devlink_health_report at ffffffff8aa9ae19
qed_report_fatal_error at ffffffffc0697baf [qed]
qed_hw_err_notify at ffffffffc06d32d7 [qed]
qed_spq_post at ffffffffc06b1011 [qed]
qed_fcoe_destroy_conn at ffffffffc06b2e91 [qed]
qedf_cleanup_fcport at ffffffffc05e7597 [qedf]
qedf_rport_event_handler at ffffffffc05e7bf7 [qedf]
fc_rport_work at ffffffffc02da715 [libfc]
process_one_work at ffffffff8a319663
Resolve this by clamping the firmware's return value to the maximum
number of legal elements the firmware should return. |
| In the Linux kernel, the following vulnerability has been resolved:
ksmbd: smbdirect: validate data_offset and data_length field of smb_direct_data_transfer
If data_offset and data_length of smb_direct_data_transfer struct are
invalid, out of bounds issue could happen.
This patch validate data_offset and data_length field in recv_done. |
| In the Linux kernel, the following vulnerability has been resolved:
cnic: Fix use-after-free bugs in cnic_delete_task
The original code uses cancel_delayed_work() in cnic_cm_stop_bnx2x_hw(),
which does not guarantee that the delayed work item 'delete_task' has
fully completed if it was already running. Additionally, the delayed work
item is cyclic, the flush_workqueue() in cnic_cm_stop_bnx2x_hw() only
blocks and waits for work items that were already queued to the
workqueue prior to its invocation. Any work items submitted after
flush_workqueue() is called are not included in the set of tasks that the
flush operation awaits. This means that after the cyclic work items have
finished executing, a delayed work item may still exist in the workqueue.
This leads to use-after-free scenarios where the cnic_dev is deallocated
by cnic_free_dev(), while delete_task remains active and attempt to
dereference cnic_dev in cnic_delete_task().
A typical race condition is illustrated below:
CPU 0 (cleanup) | CPU 1 (delayed work callback)
cnic_netdev_event() |
cnic_stop_hw() | cnic_delete_task()
cnic_cm_stop_bnx2x_hw() | ...
cancel_delayed_work() | /* the queue_delayed_work()
flush_workqueue() | executes after flush_workqueue()*/
| queue_delayed_work()
cnic_free_dev(dev)//free | cnic_delete_task() //new instance
| dev = cp->dev; //use
Replace cancel_delayed_work() with cancel_delayed_work_sync() to ensure
that the cyclic delayed work item is properly canceled and that any
ongoing execution of the work item completes before the cnic_dev is
deallocated. Furthermore, since cancel_delayed_work_sync() uses
__flush_work(work, true) to synchronously wait for any currently
executing instance of the work item to finish, the flush_workqueue()
becomes redundant and should be removed.
This bug was identified through static analysis. To reproduce the issue
and validate the fix, I simulated the cnic PCI device in QEMU and
introduced intentional delays — such as inserting calls to ssleep()
within the cnic_delete_task() function — to increase the likelihood
of triggering the bug. |
| In the Linux kernel, the following vulnerability has been resolved:
bpf: Propagate error from htab_lock_bucket() to userspace
In __htab_map_lookup_and_delete_batch() if htab_lock_bucket() returns
-EBUSY, it will go to next bucket. Going to next bucket may not only
skip the elements in current bucket silently, but also incur
out-of-bound memory access or expose kernel memory to userspace if
current bucket_cnt is greater than bucket_size or zero.
Fixing it by stopping batch operation and returning -EBUSY when
htab_lock_bucket() fails, and the application can retry or skip the busy
batch as needed. |
| In the Linux kernel, the following vulnerability has been resolved:
macintosh: fix possible memory leak in macio_add_one_device()
Afer commit 1fa5ae857bb1 ("driver core: get rid of struct device's
bus_id string array"), the name of device is allocated dynamically. It
needs to be freed when of_device_register() fails. Call put_device() to
give up the reference that's taken in device_initialize(), so that it
can be freed in kobject_cleanup() when the refcount hits 0.
macio device is freed in macio_release_dev(), so the kfree() can be
removed. |
| In the Linux kernel, the following vulnerability has been resolved:
crypto: af_alg - Set merge to zero early in af_alg_sendmsg
If an error causes af_alg_sendmsg to abort, ctx->merge may contain
a garbage value from the previous loop. This may then trigger a
crash on the next entry into af_alg_sendmsg when it attempts to do
a merge that can't be done.
Fix this by setting ctx->merge to zero near the start of the loop. |
| In the Linux kernel, the following vulnerability has been resolved:
memory: pl353-smc: Fix refcount leak bug in pl353_smc_probe()
The break of for_each_available_child_of_node() needs a
corresponding of_node_put() when the reference 'child' is not
used anymore. Here we do not need to call of_node_put() in
fail path as '!match' means no break.
While the of_platform_device_create() will created a new
reference by 'child' but it has considered the refcounting. |
| In the Linux kernel, the following vulnerability has been resolved:
cpufreq: Init completion before kobject_init_and_add()
In cpufreq_policy_alloc(), it will call uninitialed completion in
cpufreq_sysfs_release() when kobject_init_and_add() fails. And
that will cause a crash such as the following page fault in complete:
BUG: unable to handle page fault for address: fffffffffffffff8
[..]
RIP: 0010:complete+0x98/0x1f0
[..]
Call Trace:
kobject_put+0x1be/0x4c0
cpufreq_online.cold+0xee/0x1fd
cpufreq_add_dev+0x183/0x1e0
subsys_interface_register+0x3f5/0x4e0
cpufreq_register_driver+0x3b7/0x670
acpi_cpufreq_init+0x56c/0x1000 [acpi_cpufreq]
do_one_initcall+0x13d/0x780
do_init_module+0x1c3/0x630
load_module+0x6e67/0x73b0
__do_sys_finit_module+0x181/0x240
do_syscall_64+0x35/0x80
entry_SYSCALL_64_after_hwframe+0x63/0xcd |
| In the Linux kernel, the following vulnerability has been resolved:
drm/mipi-dsi: Detach devices when removing the host
Whenever the MIPI-DSI host is unregistered, the code of
mipi_dsi_host_unregister() loops over every device currently found on that
bus and will unregister it.
However, it doesn't detach it from the bus first, which leads to all kind
of resource leaks if the host wants to perform some clean up whenever a
device is detached. |
| In the Linux kernel, the following vulnerability has been resolved:
smb: client: let smbd_destroy() call disable_work_sync(&info->post_send_credits_work)
In smbd_destroy() we may destroy the memory so we better
wait until post_send_credits_work is no longer pending
and will never be started again.
I actually just hit the case using rxe:
WARNING: CPU: 0 PID: 138 at drivers/infiniband/sw/rxe/rxe_verbs.c:1032 rxe_post_recv+0x1ee/0x480 [rdma_rxe]
...
[ 5305.686979] [ T138] smbd_post_recv+0x445/0xc10 [cifs]
[ 5305.687135] [ T138] ? srso_alias_return_thunk+0x5/0xfbef5
[ 5305.687149] [ T138] ? __kasan_check_write+0x14/0x30
[ 5305.687185] [ T138] ? __pfx_smbd_post_recv+0x10/0x10 [cifs]
[ 5305.687329] [ T138] ? __pfx__raw_spin_lock_irqsave+0x10/0x10
[ 5305.687356] [ T138] ? srso_alias_return_thunk+0x5/0xfbef5
[ 5305.687368] [ T138] ? srso_alias_return_thunk+0x5/0xfbef5
[ 5305.687378] [ T138] ? _raw_spin_unlock_irqrestore+0x11/0x60
[ 5305.687389] [ T138] ? srso_alias_return_thunk+0x5/0xfbef5
[ 5305.687399] [ T138] ? get_receive_buffer+0x168/0x210 [cifs]
[ 5305.687555] [ T138] smbd_post_send_credits+0x382/0x4b0 [cifs]
[ 5305.687701] [ T138] ? __pfx_smbd_post_send_credits+0x10/0x10 [cifs]
[ 5305.687855] [ T138] ? __pfx___schedule+0x10/0x10
[ 5305.687865] [ T138] ? __pfx__raw_spin_lock_irq+0x10/0x10
[ 5305.687875] [ T138] ? queue_delayed_work_on+0x8e/0xa0
[ 5305.687889] [ T138] process_one_work+0x629/0xf80
[ 5305.687908] [ T138] ? srso_alias_return_thunk+0x5/0xfbef5
[ 5305.687917] [ T138] ? __kasan_check_write+0x14/0x30
[ 5305.687933] [ T138] worker_thread+0x87f/0x1570
...
It means rxe_post_recv was called after rdma_destroy_qp().
This happened because put_receive_buffer() was triggered
by ib_drain_qp() and called:
queue_work(info->workqueue, &info->post_send_credits_work); |
| In the Linux kernel, the following vulnerability has been resolved:
ASoC: qcom: q6apm-lpass-dais: Fix NULL pointer dereference if source graph failed
If earlier opening of source graph fails (e.g. ADSP rejects due to
incorrect audioreach topology), the graph is closed and
"dai_data->graph[dai->id]" is assigned NULL. Preparing the DAI for sink
graph continues though and next call to q6apm_lpass_dai_prepare()
receives dai_data->graph[dai->id]=NULL leading to NULL pointer
exception:
qcom-apm gprsvc:service:2:1: Error (1) Processing 0x01001002 cmd
qcom-apm gprsvc:service:2:1: DSP returned error[1001002] 1
q6apm-lpass-dais 30000000.remoteproc:glink-edge:gpr:service@1:bedais: fail to start APM port 78
q6apm-lpass-dais 30000000.remoteproc:glink-edge:gpr:service@1:bedais: ASoC: error at snd_soc_pcm_dai_prepare on TX_CODEC_DMA_TX_3: -22
Unable to handle kernel NULL pointer dereference at virtual address 00000000000000a8
...
Call trace:
q6apm_graph_media_format_pcm+0x48/0x120 (P)
q6apm_lpass_dai_prepare+0x110/0x1b4
snd_soc_pcm_dai_prepare+0x74/0x108
__soc_pcm_prepare+0x44/0x160
dpcm_be_dai_prepare+0x124/0x1c0 |
| In the Linux kernel, the following vulnerability has been resolved:
xen/gntdev: Accommodate VMA splitting
Prior to this commit, the gntdev driver code did not handle the
following scenario correctly with paravirtualized (PV) Xen domains:
* User process sets up a gntdev mapping composed of two grant mappings
(i.e., two pages shared by another Xen domain).
* User process munmap()s one of the pages.
* User process munmap()s the remaining page.
* User process exits.
In the scenario above, the user process would cause the kernel to log
the following messages in dmesg for the first munmap(), and the second
munmap() call would result in similar log messages:
BUG: Bad page map in process doublemap.test pte:... pmd:...
page:0000000057c97bff refcount:1 mapcount:-1 \
mapping:0000000000000000 index:0x0 pfn:...
...
page dumped because: bad pte
...
file:gntdev fault:0x0 mmap:gntdev_mmap [xen_gntdev] readpage:0x0
...
Call Trace:
<TASK>
dump_stack_lvl+0x46/0x5e
print_bad_pte.cold+0x66/0xb6
unmap_page_range+0x7e5/0xdc0
unmap_vmas+0x78/0xf0
unmap_region+0xa8/0x110
__do_munmap+0x1ea/0x4e0
__vm_munmap+0x75/0x120
__x64_sys_munmap+0x28/0x40
do_syscall_64+0x38/0x90
entry_SYSCALL_64_after_hwframe+0x61/0xcb
...
For each munmap() call, the Xen hypervisor (if built with CONFIG_DEBUG)
would print out the following and trigger a general protection fault in
the affected Xen PV domain:
(XEN) d0v... Attempt to implicitly unmap d0's grant PTE ...
(XEN) d0v... Attempt to implicitly unmap d0's grant PTE ...
As of this writing, gntdev_grant_map structure's vma field (referred to
as map->vma below) is mainly used for checking the start and end
addresses of mappings. However, with split VMAs, these may change, and
there could be more than one VMA associated with a gntdev mapping.
Hence, remove the use of map->vma and rely on map->pages_vm_start for
the original start address and on (map->count << PAGE_SHIFT) for the
original mapping size. Let the invalidate() and find_special_page()
hooks use these.
Also, given that there can be multiple VMAs associated with a gntdev
mapping, move the "mmu_interval_notifier_remove(&map->notifier)" call to
the end of gntdev_put_map, so that the MMU notifier is only removed
after the closing of the last remaining VMA.
Finally, use an atomic to prevent inadvertent gntdev mapping re-use,
instead of using the map->live_grants atomic counter and/or the map->vma
pointer (the latter of which is now removed). This prevents the
userspace from mmap()'ing (with MAP_FIXED) a gntdev mapping over the
same address range as a previously set up gntdev mapping. This scenario
can be summarized with the following call-trace, which was valid prior
to this commit:
mmap
gntdev_mmap
mmap (repeat mmap with MAP_FIXED over the same address range)
gntdev_invalidate
unmap_grant_pages (sets 'being_removed' entries to true)
gnttab_unmap_refs_async
unmap_single_vma
gntdev_mmap (maps the shared pages again)
munmap
gntdev_invalidate
unmap_grant_pages
(no-op because 'being_removed' entries are true)
unmap_single_vma (For PV domains, Xen reports that a granted page
is being unmapped and triggers a general protection fault in the
affected domain, if Xen was built with CONFIG_DEBUG)
The fix for this last scenario could be worth its own commit, but we
opted for a single commit, because removing the gntdev_grant_map
structure's vma field requires guarding the entry to gntdev_mmap(), and
the live_grants atomic counter is not sufficient on its own to prevent
the mmap() over a pre-existing mapping. |
| In the Linux kernel, the following vulnerability has been resolved:
RDMA/core: Make sure "ib_port" is valid when access sysfs node
The "ib_port" structure must be set before adding the sysfs kobject,
and reset after removing it, otherwise it may crash when accessing
the sysfs node:
Unable to handle kernel NULL pointer dereference at virtual address 0000000000000050
Mem abort info:
ESR = 0x96000006
Exception class = DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
Data abort info:
ISV = 0, ISS = 0x00000006
CM = 0, WnR = 0
user pgtable: 4k pages, 48-bit VAs, pgdp = 00000000e85f5ba5
[0000000000000050] pgd=0000000848fd9003, pud=000000085b387003, pmd=0000000000000000
Internal error: Oops: 96000006 [#2] PREEMPT SMP
Modules linked in: ib_umad(O) mlx5_ib(O) nfnetlink_cttimeout(E) nfnetlink(E) act_gact(E) cls_flower(E) sch_ingress(E) openvswitch(E) nsh(E) nf_nat_ipv6(E) nf_nat_ipv4(E) nf_conncount(E) nf_nat(E) nf_conntrack(E) nf_defrag_ipv6(E) nf_defrag_ipv4(E) mst_pciconf(O) ipmi_devintf(E) ipmi_msghandler(E) ipmb_dev_int(OE) mlx5_core(O) mlxfw(O) mlxdevm(O) auxiliary(O) ib_uverbs(O) ib_core(O) mlx_compat(O) psample(E) sbsa_gwdt(E) uio_pdrv_genirq(E) uio(E) mlxbf_pmc(OE) mlxbf_gige(OE) mlxbf_tmfifo(OE) gpio_mlxbf2(OE) pwr_mlxbf(OE) mlx_trio(OE) i2c_mlxbf(OE) mlx_bootctl(OE) bluefield_edac(OE) knem(O) ip_tables(E) ipv6(E) crc_ccitt(E) [last unloaded: mst_pci]
Process grep (pid: 3372, stack limit = 0x0000000022055c92)
CPU: 5 PID: 3372 Comm: grep Tainted: G D OE 4.19.161-mlnx.47.gadcd9e3 #1
Hardware name: https://www.mellanox.com BlueField SoC/BlueField SoC, BIOS BlueField:3.9.2-15-ga2403ab Sep 8 2022
pstate: 40000005 (nZcv daif -PAN -UAO)
pc : hw_stat_port_show+0x4c/0x80 [ib_core]
lr : port_attr_show+0x40/0x58 [ib_core]
sp : ffff000029f43b50
x29: ffff000029f43b50 x28: 0000000019375000
x27: ffff8007b821a540 x26: ffff000029f43e30
x25: 0000000000008000 x24: ffff000000eaa958
x23: 0000000000001000 x22: ffff8007a4ce3000
x21: ffff8007baff8000 x20: ffff8007b9066ac0
x19: ffff8007bae97578 x18: 0000000000000000
x17: 0000000000000000 x16: 0000000000000000
x15: 0000000000000000 x14: 0000000000000000
x13: 0000000000000000 x12: 0000000000000000
x11: 0000000000000000 x10: 0000000000000000
x9 : 0000000000000000 x8 : ffff8007a4ce4000
x7 : 0000000000000000 x6 : 000000000000003f
x5 : ffff000000e6a280 x4 : ffff8007a4ce3000
x3 : 0000000000000000 x2 : aaaaaaaaaaaaaaab
x1 : ffff8007b9066a10 x0 : ffff8007baff8000
Call trace:
hw_stat_port_show+0x4c/0x80 [ib_core]
port_attr_show+0x40/0x58 [ib_core]
sysfs_kf_seq_show+0x8c/0x150
kernfs_seq_show+0x44/0x50
seq_read+0x1b4/0x45c
kernfs_fop_read+0x148/0x1d8
__vfs_read+0x58/0x180
vfs_read+0x94/0x154
ksys_read+0x68/0xd8
__arm64_sys_read+0x28/0x34
el0_svc_common+0x88/0x18c
el0_svc_handler+0x78/0x94
el0_svc+0x8/0xe8
Code: f2955562 aa1603e4 aa1503e0 f9405683 (f9402861) |
| In the Linux kernel, the following vulnerability has been resolved:
drm/amd: fix potential memory leak
This patch fix potential memory leak (clk_src) when function run
into last return NULL.
s/free/kfree/ - Alex |
| In the Linux kernel, the following vulnerability has been resolved:
coresight: cti: Fix hang in cti_disable_hw()
cti_enable_hw() and cti_disable_hw() are called from an atomic context
so shouldn't use runtime PM because it can result in a sleep when
communicating with firmware.
Since commit 3c6656337852 ("Revert "firmware: arm_scmi: Add clock
management to the SCMI power domain""), this causes a hang on Juno when
running the Perf Coresight tests or running this command:
perf record -e cs_etm//u -- ls
This was also missed until the revert commit because pm_runtime_put()
was called with the wrong device until commit 692c9a499b28 ("coresight:
cti: Correct the parameter for pm_runtime_put")
With lock and scheduler debugging enabled the following is output:
coresight cti_sys0: cti_enable_hw -- dev:cti_sys0 parent: 20020000.cti
BUG: sleeping function called from invalid context at drivers/base/power/runtime.c:1151
in_atomic(): 1, irqs_disabled(): 128, non_block: 0, pid: 330, name: perf-exec
preempt_count: 2, expected: 0
RCU nest depth: 0, expected: 0
INFO: lockdep is turned off.
irq event stamp: 0
hardirqs last enabled at (0): [<0000000000000000>] 0x0
hardirqs last disabled at (0): [<ffff80000822b394>] copy_process+0xa0c/0x1948
softirqs last enabled at (0): [<ffff80000822b394>] copy_process+0xa0c/0x1948
softirqs last disabled at (0): [<0000000000000000>] 0x0
CPU: 3 PID: 330 Comm: perf-exec Not tainted 6.0.0-00053-g042116d99298 #7
Hardware name: ARM LTD ARM Juno Development Platform/ARM Juno Development Platform, BIOS EDK II Sep 13 2022
Call trace:
dump_backtrace+0x134/0x140
show_stack+0x20/0x58
dump_stack_lvl+0x8c/0xb8
dump_stack+0x18/0x34
__might_resched+0x180/0x228
__might_sleep+0x50/0x88
__pm_runtime_resume+0xac/0xb0
cti_enable+0x44/0x120
coresight_control_assoc_ectdev+0xc0/0x150
coresight_enable_path+0xb4/0x288
etm_event_start+0x138/0x170
etm_event_add+0x48/0x70
event_sched_in.isra.122+0xb4/0x280
merge_sched_in+0x1fc/0x3d0
visit_groups_merge.constprop.137+0x16c/0x4b0
ctx_sched_in+0x114/0x1f0
perf_event_sched_in+0x60/0x90
ctx_resched+0x68/0xb0
perf_event_exec+0x138/0x508
begin_new_exec+0x52c/0xd40
load_elf_binary+0x6b8/0x17d0
bprm_execve+0x360/0x7f8
do_execveat_common.isra.47+0x218/0x238
__arm64_sys_execve+0x48/0x60
invoke_syscall+0x4c/0x110
el0_svc_common.constprop.4+0xfc/0x120
do_el0_svc+0x34/0xc0
el0_svc+0x40/0x98
el0t_64_sync_handler+0x98/0xc0
el0t_64_sync+0x170/0x174
Fix the issue by removing the runtime PM calls completely. They are not
needed here because it must have already been done when building the
path for a trace.
[ Fix build warnings ] |
| In the Linux kernel, the following vulnerability has been resolved:
net: enetc: avoid buffer leaks on xdp_do_redirect() failure
Before enetc_clean_rx_ring_xdp() calls xdp_do_redirect(), each software
BD in the RX ring between index orig_i and i can have one of 2 refcount
values on its page.
We are the owner of the current buffer that is being processed, so the
refcount will be at least 1.
If the current owner of the buffer at the diametrically opposed index
in the RX ring (i.o.w, the other half of this page) has not yet called
kfree(), this page's refcount could even be 2.
enetc_page_reusable() in enetc_flip_rx_buff() tests for the page
refcount against 1, and [ if it's 2 ] does not attempt to reuse it.
But if enetc_flip_rx_buff() is put after the xdp_do_redirect() call,
the page refcount can have one of 3 values. It can also be 0, if there
is no owner of the other page half, and xdp_do_redirect() for this
buffer ran so far that it triggered a flush of the devmap/cpumap bulk
queue, and the consumers of those bulk queues also freed the buffer,
all by the time xdp_do_redirect() returns the execution back to enetc.
This is the reason why enetc_flip_rx_buff() is called before
xdp_do_redirect(), but there is a big flaw with that reasoning:
enetc_flip_rx_buff() will set rx_swbd->page = NULL on both sides of the
enetc_page_reusable() branch, and if xdp_do_redirect() returns an error,
we call enetc_xdp_free(), which does not deal gracefully with that.
In fact, what happens is quite special. The page refcounts start as 1.
enetc_flip_rx_buff() figures they're reusable, transfers these
rx_swbd->page pointers to a different rx_swbd in enetc_reuse_page(), and
bumps the refcount to 2. When xdp_do_redirect() later returns an error,
we call the no-op enetc_xdp_free(), but we still haven't lost the
reference to that page. A copy of it is still at rx_ring->next_to_alloc,
but that has refcount 2 (and there are no concurrent owners of it in
flight, to drop the refcount). What really kills the system is when
we'll flip the rx_swbd->page the second time around. With an updated
refcount of 2, the page will not be reusable and we'll really leak it.
Then enetc_new_page() will have to allocate more pages, which will then
eventually leak again on further errors from xdp_do_redirect().
The problem, summarized, is that we zeroize rx_swbd->page before we're
completely done with it, and this makes it impossible for the error path
to do something with it.
Since the packet is potentially multi-buffer and therefore the
rx_swbd->page is potentially an array, manual passing of the old
pointers between enetc_flip_rx_buff() and enetc_xdp_free() is a bit
difficult.
For the sake of going with a simple solution, we accept the possibility
of racing with xdp_do_redirect(), and we move the flip procedure to
execute only on the redirect success path. By racing, I mean that the
page may be deemed as not reusable by enetc (having a refcount of 0),
but there will be no leak in that case, either.
Once we accept that, we have something better to do with buffers on
XDP_REDIRECT failure. Since we haven't performed half-page flipping yet,
we won't, either (and this way, we can avoid enetc_xdp_free()
completely, which gives the entire page to the slab allocator).
Instead, we'll call enetc_xdp_drop(), which will recycle this half of
the buffer back to the RX ring. |
| In the Linux kernel, the following vulnerability has been resolved:
smb: client: fix smbdirect_recv_io leak in smbd_negotiate() error path
During tests of another unrelated patch I was able to trigger this
error: Objects remaining on __kmem_cache_shutdown() |
