- 22 Jun, 2022 31 commits
-
-
Yupeng Li authored
[ Upstream commit 41e45640 ] set cpu_hwmon as a module build with loongson_sysconf, loongson_chiptemp undefined error,fix cpu_hwmon compile options to be bool.Some kernel compilation error information is as follows: Checking missing-syscalls for N32 CALL scripts/checksyscalls.sh Checking missing-syscalls for O32 CALL scripts/checksyscalls.sh CALL scripts/checksyscalls.sh CHK include/generated/compile.h CC [M] drivers/platform/mips/cpu_hwmon.o Building modules, stage 2. MODPOST 200 modules ERROR: "loongson_sysconf" [drivers/platform/mips/cpu_hwmon.ko] undefined! ERROR: "loongson_chiptemp" [drivers/platform/mips/cpu_hwmon.ko] undefined! make[1]: *** [scripts/Makefile.modpost:92：__modpost] 错误 1 make: *** [Makefile:1261：modules] 错误 2 Signed-off-by:
Yupeng Li <liyupeng@zbhlos.com> Reviewed-by:
Guenter Roeck <linux@roeck-us.net> Reviewed-by:
Huacai Chen <chenhuacai@kernel.org> Signed-off-by:
Thomas Bogendoerfer <tsbogend@alpha.franken.de> Signed-off-by:
Sasha Levin <sashal@kernel.org>
-
Linus Torvalds authored
[ Upstream commit 842c3b3d ] gcc-12 started warning about 'tracker' being used uninitialized: drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c: In function ‘mlx5_do_bond’: drivers/net/ethernet/mellanox/mlx5/core/lag/lag.c:786:28: warning: ‘tracker’ is used uninitialized [-Wuninitialized] 786 | struct lag_tracker tracker; | ^~~~~~~ which seems to be because it doesn't track how the use (and initialization) is bound by the 'do_bond' flag. But admittedly that 'do_bond' usage is fairly complicated, and involves passing it around as an argument to helper functions, so it's somewhat understandable that gcc doesn't see how that all works. This function could be rewritten to make the use of that tracker variable more obviously safe, but for now I'm just adding the forced initialization of it. Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by:
-
Chen Lin authored
[ Upstream commit 2f2c0d29 ] When rx_flag == MTK_RX_FLAGS_HWLRO, rx_data_len = MTK_MAX_LRO_RX_LENGTH(4096 * 3) > PAGE_SIZE. netdev_alloc_frag is for alloction of page fragment only. Reference to other drivers and Documentation/vm/page_frags.rst Branch to use __get_free_pages when ring->frag_size > PAGE_SIZE. Signed-off-by:
Chen Lin <chen45464546@163.com> Link: https://lore.kernel.org/r/1654692413-2598-1-git-send-email-chen45464546@163.com Signed-off-by:
Jakub Kicinski <kuba@kernel.org> Signed-off-by:
-
Wang Yufen authored
[ Upstream commit f638a84a ] When len >= INT_MAX - transhdrlen, ulen = len + transhdrlen will be overflow. To fix, we can follow what udpv6 does and subtract the transhdrlen from the max. Signed-off-by:
Wang Yufen <wangyufen@huawei.com> Link: https://lore.kernel.org/r/20220607120028.845916-2-wangyufen@huawei.com Signed-off-by:
-
Xiaohui Zhang authored
[ Upstream commit 8a4d4807 ] Similar to the handling of play_deferred in commit 19cfe912 ("Bluetooth: btusb: Fix memory leak in play_deferred"), we thought a patch might be needed here as well. Currently usb_submit_urb is called directly to submit deferred tx urbs after unanchor them. So the usb_giveback_urb_bh would failed to unref it in usb_unanchor_urb and cause memory leak. Put those urbs in tx_anchor to avoid the leak, and also fix the error handling. Signed-off-by:
Xiaohui Zhang <xiaohuizhang@ruc.edu.cn> Acked-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Link: https://lore.kernel.org/r/20220607083230.6182-1-xiaohuizhang@ruc.edu.cn Signed-off-by:
-
chengkaitao authored
[ Upstream commit a58a7f97 ] The reference must be released when device_register(&vm_cmdline_parent) failed. Add the corresponding 'put_device()' in the error handling path. Signed-off-by:
chengkaitao <pilgrimtao@gmail.com> Message-Id: <20220602005542.16489-1-chengkaitao@didiglobal.com> Signed-off-by:
Michael S. Tsirkin <mst@redhat.com> Acked-by:
Jason Wang <jasowang@redhat.com> Signed-off-by:
-
huangwenhui authored
[ Upstream commit 527f4643 ] Added the support of new Huawei codec HW8326. The HW8326 is developed by Huawei with Realtek's IP Core, and it's compatible with ALC256. Signed-off-by:
huangwenhui <huangwenhuia@uniontech.com> Link: https://lore.kernel.org/r/20220608082357.26898-1-huangwenhuia@uniontech.com Signed-off-by:
Takashi Iwai <tiwai@suse.de> Signed-off-by:
-
Chengguang Xu authored
[ Upstream commit ec1e8adc ] Fix missing resource cleanup (when '(--i) == 0') for error case in pmcraid_register_interrupt_handler(). Link: https://lore.kernel.org/r/20220529153456.4183738-6-cgxu519@mykernel.net Reviewed-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
Chengguang Xu <cgxu519@mykernel.net> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by:
-
Chengguang Xu authored
[ Upstream commit d64c4919 ] Fix missing resource cleanup (when '(--i) == 0') for error case in ipr_alloc_mem() and skip incorrect resource cleanup (when '(--i) == 0') for error case in ipr_request_other_msi_irqs() because variable i started from 1. Link: https://lore.kernel.org/r/20220529153456.4183738-4-cgxu519@mykernel.net Reviewed-by:
Brian King <brking@linux.vnet.ibm.com> Signed-off-by:
-
James Smart authored
[ Upstream commit 2e7e9c0c ] NVMe Asynchronous Event Request commands have no command timeout value per specifications. Set WQE option to allow a reduced FLUSH polling rate for I/O error detection specifically for nvme_admin_async_event commands. Link: https://lore.kernel.org/r/20220603174329.63777-9-jsmart2021@gmail.com Co-developed-by:
Justin Tee <justin.tee@broadcom.com> Signed-off-by:
James Smart <jsmart2021@gmail.com> Signed-off-by:
-
James Smart authored
[ Upstream commit 336d6361 ] After issuing a LIP, a specific target vendor does not ACC the FLOGI that lpfc sends. However, it does send its own FLOGI that lpfc ACCs. The target then establishes the port IDs by sending a PLOGI. lpfc PLOGI_ACCs and starts the RPI registration for DID 0x000001. The target then sends a LOGO to the fabric DID. lpfc is currently treating the LOGO from the fabric DID as a link down and cleans up all the ndlps. The ndlp for DID 0x000001 is put back into NPR and discovery stops, leaving the port in stuck in bypassed mode. Change lpfc behavior such that if a LOGO is received for the fabric DID in PT2PT topology skip the lpfc_linkdown_port() routine and just move the fabric DID back to NPR. Link: https://lore.kernel.org/r/20220603174329.63777-7-jsmart2021@gmail.com Co-developed-by:
-
Wentao Wang authored
[ Upstream commit cf71d59c ] vcpuHint has been expanded to 16 bit on host to enable routing to more CPUs. Guest side should align with the change. This change has been tested with hosts with 8-bit and 16-bit vcpuHint, on both platforms host side can get correct value. Link: https://lore.kernel.org/r/EF35F4D5-5DCC-42C5-BCC4-29DF1729B24C@vmware.com Signed-off-by:
Wentao Wang <wwentao@vmware.com> Signed-off-by:
-
Marius Hoch authored
[ Upstream commit 6ab2e518 ] Commit 223f61b8 ("Input: soc_button_array - add Lenovo Yoga Tablet2 1051L to the dmi_use_low_level_irq list") added the 1051L to this list already, but the same problem applies to the 1051F. As there are no further 1051 variants (just the F/L), we can just DMI match 1051. Tested on a Lenovo Yoga Tablet2 1051F: Without this patch the home-button stops working after a wakeup from suspend. Signed-off-by:
Marius Hoch <mail@mariushoch.de> Reviewed-by:
Hans de Goede <hdegoede@redhat.com> Link: https://lore.kernel.org/r/20220603120246.3065-1-mail@mariushoch.de Signed-off-by:
Dmitry Torokhov <dmitry.torokhov@gmail.com> Signed-off-by:
-
Mark Brown authored
[ Upstream commit 2abdf9f8 ] Currently wm_adsp_fw_put() returns 0 rather than 1 when updating the value of the control, meaning that no event is generated to userspace. Fix this by setting the default return value to 1, the code already exits early with a return value of 0 if the value is unchanged. Signed-off-by:
Mark Brown <broonie@kernel.org> Reviewed-by:
Richard Fitzgerald <rf@opensource.cirrus.com> Link: https://lore.kernel.org/r/20220603115003.3865834-1-broonie@kernel.org Signed-off-by:
-
Mark Brown authored
[ Upstream commit 8259610c ] Currently the put() method for the deemphasis control returns 0 when a new value is written to the control even if the value changed, meaning events are not generated. Fix this, skip the work of updating the value when it is unchanged and then return 1 after having done so. Signed-off-by:
-
Adam Ford authored
[ Upstream commit d1f5272c ] If the audio CODEC is playing sound when the system is suspended, it can be left in a state which throws the following error: wm8962 3-001a: ASoC: error at soc_component_read_no_lock on wm8962.3-001a: -16 Once this error has occurred, the audio will not work again until rebooted. Fix this by configuring SET_SYSTEM_SLEEP_PM_OPS. Signed-off-by:
Adam Ford <aford173@gmail.com> Acked-by:
Charles Keepax <ckeepax@opensource.cirrus.com> Link: https://lore.kernel.org/r/20220526182129.538472-1-aford173@gmail.com Signed-off-by:
-
Matthew Wilcox (Oracle) authored
[ Upstream commit 537e11cd ] As described in commit 02117b8a ("f2fs: Set GF_NOFS in read_cache_page_gfp while doing f2fs_quota_read"), we must not enter filesystem reclaim while holding the dq_lock. Prevent this more generally by using memalloc_nofs_save() while holding the lock. Link: https://lore.kernel.org/r/20220605143815.2330891-2-willy@infradead.org Signed-off-by:
Matthew Wilcox (Oracle) <willy@infradead.org> Signed-off-by:
Jan Kara <jack@suse.cz> Signed-off-by:
-
Sergey Shtylyov authored
[ Upstream commit bf476fe2 ] In an unlikely (and probably wrong?) case that the 'ppi' parameter of ata_host_alloc_pinfo() points to an array starting with a NULL pointer, there's going to be a kernel oops as the 'pi' local variable won't get reassigned from the initial value of NULL. Initialize 'pi' instead to '&ata_dummy_port_info' to fix the possible kernel oops for good... Found by Linux Verification Center (linuxtesting.org) with the SVACE static analysis tool. Signed-off-by:
Sergey Shtylyov <s.shtylyov@omp.ru> Signed-off-by:
Damien Le Moal <damien.lemoal@opensource.wdc.com> Signed-off-by:
-
Charles Keepax authored
[ Upstream commit fcb3b5a5 ] The minimum value for the PGA Volume is given as 0x1A, however the values from there to 0x19 are all the same volume and this is not represented in the TLV structure. The number of volumes given is correct so this leads to all the volumes being shifted. Move the minimum value up to 0x19 to fix this. Signed-off-by:
-
Charles Keepax authored
[ Upstream commit a8928ada ] A couple of the SX volume controls specify 0x84 as the lowest volume value, however the correct value from the datasheet is 0x44. The datasheet don't include spaces in the value it displays as binary so this was almost certainly just a typo reading 1000100. Signed-off-by:
-
Charles Keepax authored
[ Upstream commit 91e90c71 ] The Bypass Volume is accidentally using a -6dB minimum TLV rather than the correct -60dB minimum. Add a new TLV to correct this. Signed-off-by:
-
Charles Keepax authored
[ Upstream commit 7fbd6dd6 ] This driver specified the maximum value rather than the number of volume levels on the SX controls, this is incorrect, so correct them. Reported-by:
David Rhodes <david.rhodes@cirrus.com> Signed-off-by:
-
Charles Keepax authored
[ Upstream commit 5005a234 ] The digital volume TLV specifies the step as 0.25dB but the actual step of the control is 0.125dB. Update the TLV to correct this. Signed-off-by:
-
Charles Keepax authored
[ Upstream commit 8bf5aabf ] The datasheet specifies the range of the mixer volumes as between -51.5dB and 12dB with a 0.5dB step. Update the TLVs for this. Signed-off-by:
-
Rob Clark authored
[ Upstream commit e19f8fa6 ] Limit the error msg to avoid flooding the console. If you have a lot of threads hitting this at once, they could have already gotten passed the dma_debug_disabled() check before they get to the point of allocation failure, resulting in quite a lot of this error message spamming the log. Use pr_err_once() to limit that. Signed-off-by:
Rob Clark <robdclark@chromium.org> Signed-off-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Hui Wang authored
[ Upstream commit aeca8a32 ] We tried to enable the audio on an imx6sx EVB with the codec nau8822, after setting the internal PLL fractional parameters, the audio still couldn't work and the there was no sdma irq at all. After checking with the section "8.1.1 Phase Locked Loop (PLL) Design Example" of "NAU88C22 Datasheet Rev 0.6", we found we need to turn off the PLL before programming fractional parameters and turn on the PLL after programming. After this change, the audio driver could record and play sound and the sdma's irq is triggered when playing or recording. Cc: David Lin <ctlin0@nuvoton.com> Cc: John Hsu <kchsu0@nuvoton.com> Cc: Seven Li <wtli@nuvoton.com> Signed-off-by:
Hui Wang <hui.wang@canonical.com> Link: https://lore.kernel.org/r/20220530040151.95221-2-hui.wang@canonical.com Signed-off-by:
-
He Ying authored
[ Upstream commit a1b29ba2 ] The following KASAN warning was reported in our kernel. BUG: KASAN: stack-out-of-bounds in get_wchan+0x188/0x250 Read of size 4 at addr d216f958 by task ps/14437 CPU: 3 PID: 14437 Comm: ps Tainted: G O 5.10.0 #1 Call Trace: [daa63858] [c0654348] dump_stack+0x9c/0xe4 (unreliable) [daa63888] [c035cf0c] print_address_description.constprop.3+0x8c/0x570 [daa63908] [c035d6bc] kasan_report+0x1ac/0x218 [daa63948] [c00496e8] get_wchan+0x188/0x250 [daa63978] [c0461ec8] do_task_stat+0xce8/0xe60 [daa63b98] [c0455ac8] proc_single_show+0x98/0x170 [daa63bc8] [c03cab8c] seq_read_iter+0x1ec/0x900 [daa63c38] [c03cb47c] seq_read+0x1dc/0x290 [daa63d68] [c037fc94] vfs_read+0x164/0x510 [daa63ea8] [c03808e4] ksys_read+0x144/0x1d0 [daa63f38] [c005b1dc] ret_from_syscall+0x0/0x38 --- interrupt: c00 at 0x8fa8f4 LR = 0x8fa8cc The buggy address belongs to the page: page:98ebcdd2 refcount:0 mapcount:0 mapping:00000000 index:0x2 pfn:0x1216f flags: 0x0() raw: 00000000 00000000 01010122 00000000 00000002 00000000 ffffffff 00000000 raw: 00000000 page dumped because: kasan: bad access detected Memory state around the buggy address: d216f800: 00 00 00 00 00 f1 f1 f1 f1 00 00 00 00 00 00 00 d216f880: f2 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 >d216f900: 00 00 00 00 00 00 00 00 00 00 00 f1 f1 f1 f1 00 ^ d216f980: f2 f2 f2 f2 f2 f2 f2 00 00 00 00 00 00 00 00 00 d216fa00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 After looking into this issue, I find the buggy address belongs to the task stack region. It seems KASAN has something wrong. I look into the code of __get_wchan in x86 architecture and find the same issue has been resolved by the commit f7d27c35 ("x86/mm, kasan: Silence KASAN warnings in get_wchan()"). The solution could be applied to powerpc architecture too. As Andrey Ryabinin said, get_wchan() is racy by design, it may access volatile stack of running task, thus it may access redzone in a stack frame and cause KASAN to warn about this. Use READ_ONCE_NOCHECK() to silence these warnings. Reported-by:
Wanming Hu <huwanming@huaweil.com> Signed-off-by:
He Ying <heying24@huawei.com> Signed-off-by:
Chen Jingwen <chenjingwen6@huawei.com> Reviewed-by:
Kees Cook <keescook@chromium.org> Signed-off-by:
Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20220121014418.155675-1-heying24@huawei.com Signed-off-by:
-
Adam Ford authored
commit 4ce01ce3 upstream. There is a header for a DB9 serial port, but any attempts to use hardware handshaking fail. Enable RTS and CTS pin muxing and enable handshaking in the uart node. Signed-off-by:
Shawn Guo <shawnguo@kernel.org> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Yuntao Wang authored
commit b4504319 upstream. This is a backport of the original upstream patch for 5.4/5.10. The original upstream patch has been applied to 5.4/5.10 branches, which simply removed the line: cost += n_buckets * (value_size + sizeof(struct stack_map_bucket)); This is correct for upstream branch but incorrect for 5.4/5.10 branches, as the 5.4/5.10 branches do not have the commit 37086810 ("bpf: Eliminate rlimit-based memory accounting for stackmap maps"), so the bpf_map_charge_init() function has not been removed. Currently the bpf_map_charge_init() function in 5.4/5.10 branches takes a wrong memory charge cost, the attr->max_entries * (sizeof(struct stack_map_bucket) + (u64)value_size)) part is missing, let's fix it. Cc: <stable@vger.kernel.org> # 5.4.y Cc: <stable@vger.kernel.org> # 5.10.y Signed-off-by:
Yuntao Wang <ytcoode@gmail.com> Signed-off-by:
-
Trond Myklebust authored
commit 555dbf1a upstream. The nfsd_file nf_rwsem is currently being used to separate file write and commit instances to ensure that we catch errors and apply them to the correct write/commit. We can improve scalability at the expense of a little accuracy (some extra false positives) by replacing the nf_rwsem with more careful use of the errseq_t mechanism to track errors across the different operations. Signed-off-by:
Trond Myklebust <trond.myklebust@hammerspace.com> Signed-off-by:
Chuck Lever <chuck.lever@oracle.com> [ cel: rebased on zero-verifier fix ] Signed-off-by:
Leah Rumancik <leah.rumancik@gmail.com> Signed-off-by:
-
Al Viro authored
commit b577d0cd upstream. In commit 45089142 Aneesh had missed one (admittedly, very unlikely to hit) case in v9fs_stat2inode_dotl(). However, the same considerations apply there as well - we have no business whatsoever to change ->i_rdev or the file type. Cc: Tadeusz Struk <tadeusz.struk@linaro.org> Signed-off-by:
Al Viro <viro@zeniv.linux.org.uk> Signed-off-by:
-
- 16 Jun, 2022 9 commits
-
-
Greg Kroah-Hartman authored
Link: https://lore.kernel.org/r/20220614183719.878453780@linuxfoundation.org Tested-by:
Florian Fainelli <f.fainelli@gmail.com> Tested-by:
Shuah Khan <skhan@linuxfoundation.org> Tested-by:
Fox Chen <foxhlchen@gmail.com> Tested-by:
Sudip Mukherjee <sudip.mukherjee@codethink.co.uk> Tested-by:
Linux Kernel Functional Testing <lkft@linaro.org> Tested-by:
Hulk Robot <hulkrobot@huawei.com> Tested-by:
Tyler Hicks <tyhicks@linux.microsoft.com> Tested-by:
Salvatore Bonaccorso <carnil@debian.org> Tested-by:
Jon Hunter <jonathanh@nvidia.com> Tested-by:
Pavel Machek (CIP) <pavel@denx.de> Signed-off-by:
-
Josh Poimboeuf authored
commit 1dc6ff02 upstream Similar to MDS and TAA, print a warning if SMT is enabled for the MMIO Stale Data vulnerability. Signed-off-by:
Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by:
Thomas Gleixner <tglx@linutronix.de> Signed-off-by:
-
Pawan Gupta authored
commit 027bbb88 upstream The enumeration of MD_CLEAR in CPUID(EAX=7,ECX=0).EDX{bit 10} is not an accurate indicator on all CPUs of whether the VERW instruction will overwrite fill buffers. FB_CLEAR enumeration in IA32_ARCH_CAPABILITIES{bit 17} covers the case of CPUs that are not vulnerable to MDS/TAA, indicating that microcode does overwrite fill buffers. Guests running in VMM environments may not be aware of all the capabilities/vulnerabilities of the host CPU. Specifically, a guest may apply MDS/TAA mitigations when a virtual CPU is enumerated as vulnerable to MDS/TAA even when the physical CPU is not. On CPUs that enumerate FB_CLEAR_CTRL the VMM may set FB_CLEAR_DIS to skip overwriting of fill buffers by the VERW instruction. This is done by setting FB_CLEAR_DIS during VMENTER and resetting on VMEXIT. For guests that enumerate FB_CLEAR (explicitly asking for fill buffer clear capability) the VMM will not use FB_CLEAR_DIS. Irrespective of guest state, host overwrites CPU buffers before VMENTER to protect itself from an MMIO capable guest, as part of mitigation for MMIO Stale Data vulnerabilities. Signed-off-by:
Pawan Gupta <pawan.kumar.gupta@linux.intel.com> Signed-off-by:
Borislav Petkov <bp@suse.de> Signed-off-by:
-
Pawan Gupta authored
commit a992b8a4 upstream The Shared Buffers Data Sampling (SBDS) variant of Processor MMIO Stale Data vulnerabilities may expose RDRAND, RDSEED and SGX EGETKEY data. Mitigation for this is added by a microcode update. As some of the implications of SBDS are similar to SRBDS, SRBDS mitigation infrastructure can be leveraged by SBDS. Set X86_BUG_SRBDS and use SRBDS mitigation. Mitigation is enabled by default; use srbds=off to opt-out. Mitigation status can be checked from below file: /sys/devices/system/cpu/vulnerabilities/srbds Signed-off-by:
-
Pawan Gupta authored
commit 22cac9c6 upstream Currently, Linux disables SRBDS mitigation on CPUs not affected by MDS and have the TSX feature disabled. On such CPUs, secrets cannot be extracted from CPU fill buffers using MDS or TAA. Without SRBDS mitigation, Processor MMIO Stale Data vulnerabilities can be used to extract RDRAND, RDSEED, and EGETKEY data. Do not disable SRBDS mitigation by default when CPU is also affected by Processor MMIO Stale Data vulnerabilities. Signed-off-by:
-
Pawan Gupta authored
commit 8d50cdf8 upstream Add the sysfs reporting file for Processor MMIO Stale Data vulnerability. It exposes the vulnerability and mitigation state similar to the existing files for the other hardware vulnerabilities. Signed-off-by:
-
Pawan Gupta authored
commit 99a83db5 upstream When the CPU is affected by Processor MMIO Stale Data vulnerabilities, Fill Buffer Stale Data Propagator (FBSDP) can propagate stale data out of Fill buffer to uncore buffer when CPU goes idle. Stale data can then be exploited with other variants using MMIO operations. Mitigate it by clearing the Fill buffer before entering idle state. Signed-off-by:
Josh Poimboeuf <jpoimboe@redhat.com> Signed-off-by:
-
Pawan Gupta authored
commit e5925fb8 upstream MDS, TAA and Processor MMIO Stale Data mitigations rely on clearing CPU buffers. Moreover, status of these mitigations affects each other. During boot, it is important to maintain the order in which these mitigations are selected. This is especially true for md_clear_update_mitigation() that needs to be called after MDS, TAA and Processor MMIO Stale Data mitigation selection is done. Introduce md_clear_select_mitigation(), and select all these mitigations from there. This reflects relationships between these mitigations and ensures proper ordering. Signed-off-by:
-
Pawan Gupta authored
commit 8cb861e9 upstream Processor MMIO Stale Data is a class of vulnerabilities that may expose data after an MMIO operation. For details please refer to Documentation/admin-guide/hw-vuln/processor_mmio_stale_data.rst. These vulnerabilities are broadly categorized as: Device Register Partial Write (DRPW): Some endpoint MMIO registers incorrectly handle writes that are smaller than the register size. Instead of aborting the write or only copying the correct subset of bytes (for example, 2 bytes for a 2-byte write), more bytes than specified by the write transaction may be written to the register. On some processors, this may expose stale data from the fill buffers of the core that created the write transaction. Shared Buffers Data Sampling (SBDS): After propagators may have moved data around the uncore and copied stale data into client core fill buffers, processors affected by MFBDS can leak data from the fill buffer. Shared Buffers Data Read (SBDR): It is similar to Shared Buffer Data Sampling (SBDS) except that the data is directly read into the architectural software-visible state. An attacker can use these vulnerabilities to extract data from CPU fill buffers using MDS and TAA methods. Mitigate it by clearing the CPU fill buffers using the VERW instruction before returning to a user or a guest. On CPUs not affected by MDS and TAA, user application cannot sample data from CPU fill buffers using MDS or TAA. A guest with MMIO access can still use DRPW or SBDR to extract data architecturally. Mitigate it with VERW instruction to clear fill buffers before VMENTER for MMIO capable guests. Add a kernel parameter mmio_stale_data={off|full|full,nosmt} to control the mitigation. Signed-off-by:
-
