Merge ACPI resources management quirks, ACPI NUMA updates, an ACPI LPSS
(Intel SoC) driver update and an ACPI watchdog driver fixup for 6.8-rc1:
- Add IRQ override quirks for some Infinity laptops and for TongFang
GMxXGxx (David McFarland, Hans de Goede).
- Clean up the ACPI NUMA code and fix it to ensure that fake_pxm is not
the same as one of the real pxm values (Yuntao Wang).
- Fix the fractional clock divider flags in the ACPI LPSS (Intel SoC)
driver so as to prevent miscalculation of the values in the clock
divider (Andy Shevchenko).
- Adjust comments in the ACPI watchdog driver to prevent kernel-doc
from complaining during documentation builds (Randy Dunlap).
* acpi-resource:
ACPI: resource: Add Infinity laptops to irq1_edge_low_force_override
ACPI: resource: Add another DMI match for the TongFang GMxXGxx
* acpi-numa:
ACPI: NUMA: Fix the logic of getting the fake_pxm value
ACPI: NUMA: Optimize the check for the availability of node values
ACPI: NUMA: Remove unnecessary check in acpi_parse_gi_affinity()
* acpi-soc:
ACPI: LPSS: Fix the fractional clock divider flags
* acpi-misc:
ACPI: watchdog: fix kernel-doc warnings
Merge an ACPI power management change, ACPI backlight driver changes, APEI
updates and ACPI extlog driver changes for 6.8-rc1:
- Modify the ACPI LPIT table handling code to avoid u32 multiplication
overflows in state residency computations (Nikita Kiryushin).
- Drop an unused helper function from the ACPI backlight (video) driver
and add a clarifying comment to it (Hans de Goede).
- Update the ACPI backlight driver to avoid using uninitialized memory
in some cases (Nikita Kiryushin).
- Add ACPI backlight quirk for the Colorful X15 AT 23 laptop (Yuluo
Qiu).
- Add support for vendor-defined error types to the ACPI APEI error
injection code (Avadhut Naik).
- Adjust APEI to properly set MF_ACTION_REQUIRED on synchronous memory
failure events, so they are handled differently from the asynchronous
ones (Shuai Xue).
- Fix NULL pointer dereference check in the ACPI extlog driver (Prarit
Bhargava).
- Adjust the ACPI extlog driver to clear the Extended Error Log status
when RAS_CEC handled the error (Tony Luck).
* acpi-pm:
ACPI: LPIT: Avoid u32 multiplication overflow
* acpi-video:
ACPI: video: Add quirk for the Colorful X15 AT 23 Laptop
ACPI: video: check for error while searching for backlight device parent
ACPI: video: Drop should_check_lcd_flag()
ACPI: video: Add comment about acpi_video_backlight_use_native() usage
* acpi-apei:
ACPI: APEI: set memory failure flags as MF_ACTION_REQUIRED on synchronous events
ACPI: APEI: EINJ: Add support for vendor defined error types
platform/chrome: cros_ec_debugfs: Fix permissions for panicinfo
fs: debugfs: Add write functionality to debugfs blobs
ACPI: APEI: EINJ: Refactor available_error_type_show()
* acpi-extlog:
ACPI: extlog: Clear Extended Error Log status when RAS_CEC handled the error
ACPI: extlog: fix NULL pointer dereference check
Merge ACPI thermal zone driver updates for 6.8-rc1:
- Use generic ACPI helpers for evaluating trip point temperature
objects in the ACPI thermal zone driver (Rafael J. Wysockii, Arnd
Bergmann).
- Add Thermal fast Sampling Period (_TFP) support to the ACPI thermal
zone driver (Jeff Brasen).
* acpi-thermal:
ACPI: thermal_lib: include "internal.h" for function prototypes
ACPI: thermal: Add Thermal fast Sampling Period (_TFP) support
ACPI: thermal: Use library functions to obtain trip point temperature values
ACPI: thermal_lib: Add functions returning temperature in deci-Kelvin
thermal: ACPI: Move the ACPI thermal library to drivers/acpi/
Merge ACPI utility functions updates for 6.8-rc1:
- Modify acpi_dev_uid_match() to support different types of its second
argument and adjust its users accordingly (Raag Jadav).
- Clean up code related to acpi_evaluate_reference() and ACPI device
lists (Rafael J. Wysocki).
* acpi-utils:
ACPI: utils: Introduce helper for _DEP list lookup
ACPI: utils: Fix white space in struct acpi_handle_list definition
ACPI: utils: Refine acpi_handle_list_equal() slightly
ACPI: utils: Return bool from acpi_evaluate_reference()
ACPI: utils: Rearrange in acpi_evaluate_reference()
perf: arm_cspmu: drop redundant acpi_dev_uid_to_integer()
efi: dev-path-parser: use acpi_dev_uid_match() for matching _UID
ACPI: LPSS: use acpi_dev_uid_match() for matching _UID
ACPI: bus: update acpi_dev_hid_uid_match() to support multiple types
ACPI: bus: update acpi_dev_uid_match() to support multiple types
Merge low-level ACPICA interface changes, an _SB-scope _OSC handshake
update and a data-only ACPI tables parsing code update for 6.8-rc1:
- Switch over ACPI to using a threaded interrupt handler for the
SCI (Rafael J. Wysocki).
- Allow ACPI Notify () handlers to run on all CPUs and clean up the
ACPI interface for deferred events processing (Rafael J. Wysocki).
- Switch over the ACPI EC driver to using a threaded handler for the
dedicated IRQ on systems without the EC GPE (Rafael J. Wysocki).
- Adjust code using ACPICA spinlocks and the ACPI EC driver spinlock to
keep local interrupts on (Rafael J. Wysocki).
- Adjust the USB4 _OSC handshake to correctly handle cases in which
certain types of OS control are denied by the platform (Mika
Westerberg).
- Correct and clean up the generic function for parsing ACPI data-only
tables with array structure (Yuntao Wang).
* acpi-osl:
ACPI: EC: Use a spin lock without disabing interrupts
ACPI: EC: Use a threaded handler for dedicated IRQ
ACPI: OSL: Use spin locks without disabling interrupts
ACPI: OSL: Allow Notify () handlers to run on all CPUs
ACPI: OSL: Rearrange workqueue selection in acpi_os_execute()
ACPI: OSL: Rework error handling in acpi_os_execute()
ACPI: OSL: Use a threaded interrupt handler for SCI
* acpi-bus:
ACPI: Run USB4 _OSC() first with query bit set
* acpi-tables:
ACPI: tables: Correct and clean up the logic of acpi_parse_entries_array()
Merge ACPI device enumeration updates and ACPI processor driver updates
for 6.8-rc1:
- Add CSI-2 and DisCo for Imaging support to the ACPI device
enumeration code (Sakari Ailus, Rafael J. Wysocki).
- Adjust the cpufreq thermal reduction algorithm in the ACPI processor
driver for Tegra241 (Srikar Srimath Tirumala, Arnd Bergmann).
- Make acpi_proc_quirk_mwait_check() x86-specific (Rafael J. Wysocki).
* acpi-scan:
ACPI: scan: Fix an error message in DisCo for Imaging support
ACPI: property: Replicate DT-aligned u32 properties from DisCo for Imaging
ACPI: property: Dig "rotation" property for devices with CSI2 _CRS
ACPI: scan: Extract MIPI DisCo for Imaging data into swnodes
device property: Add SOFTWARE_NODE() macro for defining software nodes
ACPI: scan: Extract _CRS CSI-2 connection information into swnodes
ACPI: scan: Extract CSI-2 connection graph from _CRS
ACPI: property: Support using strings in reference properties
* acpi-processor:
ACPI: arm64: export acpi_arch_thermal_cpufreq_pctg()
ACPI: processor: reduce CPUFREQ thermal reduction pctg for Tegra241
ACPI: processor: Provide empty stub of acpi_proc_quirk_mwait_check()
Move phylink_pcs_neg_mode() from the header file into the .c file since
nothing should be using it.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Luiz Angelo Daros de Luca reports that the MDIO bus code maintains a
reference to the DT node, but does not hold a refcount on the node.
The simple solution to this is to add the necessary refcounting into
the MDIO bus code for all users, ensuring that on registration, the
refcount is incremented, and only dropped when the MDIO bus is
released.
Do this for fwnodes, so we not only fix this for DT, but also other
types of firmware nodes as well.
Reported-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Because thermal_zone_trip_id() does not update the thermal zone object
passed to it, its pointer argument representing the thermal zone can be
const, so adjust its definition accordingly.
No functional impact.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Fix the following compile error:
.../bnxt.c: In function 'bnxt_cfg_ntp_filters':
.../bnxt.c:14077:37: error: implicit declaration of function 'rps_may_expire_flow' [-Werror=implicit-function-declaration]
14077 | if (rps_may_expire_flow(bp->dev, fltr->base.rxq,
| ^~~~~~~~~~~~~~~~~~~
bnxt_cfg_ntp_filters() is only used when CONFIG_RFS_ACCEL is enabled.
User configured ntuple filters are directly added and will not go through
this function. Wrap the body of bnxt_cfg_ntp_filters() with
CONFIG_RFS_ACCEL.
Fixes: 59cde76f33 ("bnxt_en: Refactor filter insertion logic in bnxt_rx_flow_steer().")
Reported-by: Arnd Bergmann <arnd@kernel.org>
Link: https://lore.kernel.org/netdev/20240103102332.3642417-1-arnd@kernel.org/
Reported-by: kernel test robot <lkp@intel.com>
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 2 lines to check for the BNXT_HWRM_PF_UNLOAD_SP_EVENT bit was
mis-applied to bnxt_cfg_ntp_filters() and should have been applied to
bnxt_sp_task().
Fixes: 1924136844 ("bnxt_en: Send PF driver unload notification to all VFs.")
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
CSR.OPS bits specify the current operating mode and (according to
documentation) they are updated by HW when the operating mode change
request is processed. To comply with this check CSR.OPS before proceeding.
Commit introduces ravb_set_opmode() that does all the necessities for
setting the operating mode (set CCC.OPC (and CCC.GAC, CCC.CSEL, if any) and
wait for CSR.OPS) and call it where needed. This should comply with all the
HW manuals requirements as different manual variants specify that different
modes need to be checked in CSR.OPS when setting CCC.OPC.
If gPTP active in config mode is supported and it needs to be enabled, the
CCC.GAC and CCC.CSEL needs to be configured along with CCC.OPC in the same
write access. For this, ravb_set_opmode() allows passing GAC and CSEL as
part of opmode and the function updates accordingly CCC register.
Fixes: c156633f13 ("Renesas Ethernet AVB driver proper")
Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@bp.renesas.com>
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add check for usbnet_get_endpoints() and return the error if it fails
in order to transfer the error.
Fixes: 16626b0cc3 ("asix: Add a new driver for the AX88172A")
Signed-off-by: Chen Ni <nichen@iscas.ac.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jiawen Wu says:
====================
Implement more ethtool_ops for Wangxun
Provide ethtool functions to operate pause param, ring param, coalesce
channel number and msglevel, for driver txgbe/ngbe.
v6 -> v7:
- Rebase on net-next.
v5 -> v6:
- Minor fixes address on Jakub Kicinski's comments.
v4 -> v5:
- Fix build error reported by kernel test robot.
v3 -> v4:
- Repartition the patches of phylink.
- Handle failure to allocate memory while changing ring parameters.
- Minor fixes about formatting.
v2 -> v3:
- Address comments:
https://lore.kernel.org/all/ZW2loxTO6oKNYLew@shell.armlinux.org.uk/
v1 -> v2:
- Add phylink support for ngbe.
- Fix issue on interrupts when queue number is changed.
- Add more marco defines.
- Fix return codes.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support to get RX/TX queue number with ethtool -l, and set RX/TX
queue number with ethtool -L. Since interrupts need to be rescheduled,
adjust the allocation of msix enties.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Support to show RX/TX coalesce with ethtool -c and set RX/TX
coalesce with ethtool -C.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Support to query RX/TX depth with ethtool -g, and change RX/TX depth
with ethtool -G.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support to set pause params with ethtool -A and get pause
params with ethtool -a, for ethernet driver txgbe and ngbe.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the following implementation, add struct phylink and phylink_config
to wx structure. Add the helper function for converting phylink to wx,
implement ethtool ksetting and nway reset in libwx.
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
t-queue
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2024-01-02 (ice)
This series contains updates to ice driver only.
Karol adds support for capable devices to receive timestamp via
interrupt rather than polling to allow for less delay.
Andrii adds support switchdev hardware packet mirroring.
Jake reworks VF rebuild to avoid destroying objects that do not need to
be.
Jan S removes reporting of rx_len_errors as they are incorrectly reported
by hardware.
Jan G adds const modifier to some uses that are applicable.
Kunwu Chan adds some checks for failed memory allocations.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
During QoS scheduling testing with multiple strict priority flows, the
netdev tx watchdog timeout routine is invoked when a low priority QoS
queue doesn't get a chance to transmit the packets because other high
priority flows are completely subscribing the transmit link. The netdev
tx watchdog timeout routine will stop MAC RX and TX functionality in
otx2_stop() routine before cleanup of HW TX queues which results in SMQ
flush errors because the packets belonging to low priority queues will
never gets flushed since MAC TX is disabled. This patch fixes the issue
by re-enabling MAC TX to ensure the packets in HW pipeline gets flushed
properly.
Fixes: a7faa68b4e ("octeontx2-af: Start/Stop traffic in CGX along with NPC")
Signed-off-by: Naveen Mamindlapalli <naveenm@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently the NIX TX link credits are initialized based on the max frame
size that can be transmitted on a link but when the MTU is changed, the
NIX TX link credits are reprogrammed by the SW based on the new MTU value.
Since SMQ max packet length is programmed to max frame size by default,
there is a chance that NIX TX may stall while sending a max frame sized
packet on the link with insufficient credits to send the packet all at
once. This patch avoids stall issue by not changing the link credits
dynamically when the MTU is changed.
Fixes: 1c74b89171 ("octeontx2-af: Wait for TX link idle for credits change")
Signed-off-by: Naveen Mamindlapalli <naveenm@marvell.com>
Signed-off-by: Sunil Kovvuri Goutham <sgoutham@marvell.com>
Signed-off-by: Nithin Kumar Dabilpuram <ndabilpuram@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When processing a packed profile in unpack_profile() described like
"profile :ns::samba-dcerpcd /usr/lib*/samba/{,samba/}samba-dcerpcd {...}"
a string ":samba-dcerpcd" is unpacked as a fully-qualified name and then
passed to aa_splitn_fqname().
aa_splitn_fqname() treats ":samba-dcerpcd" as only containing a namespace.
Thus it returns NULL for tmpname, meanwhile tmpns is non-NULL. Later
aa_alloc_profile() crashes as the new profile name is NULL now.
general protection fault, probably for non-canonical address 0xdffffc0000000000: 0000 [#1] PREEMPT SMP KASAN NOPTI
KASAN: null-ptr-deref in range [0x0000000000000000-0x0000000000000007]
CPU: 6 PID: 1657 Comm: apparmor_parser Not tainted 6.7.0-rc2-dirty #16
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.2-3-gd478f380-rebuilt.opensuse.org 04/01/2014
RIP: 0010:strlen+0x1e/0xa0
Call Trace:
<TASK>
? strlen+0x1e/0xa0
aa_policy_init+0x1bb/0x230
aa_alloc_profile+0xb1/0x480
unpack_profile+0x3bc/0x4960
aa_unpack+0x309/0x15e0
aa_replace_profiles+0x213/0x33c0
policy_update+0x261/0x370
profile_replace+0x20e/0x2a0
vfs_write+0x2af/0xe00
ksys_write+0x126/0x250
do_syscall_64+0x46/0xf0
entry_SYSCALL_64_after_hwframe+0x6e/0x76
</TASK>
---[ end trace 0000000000000000 ]---
RIP: 0010:strlen+0x1e/0xa0
It seems such behaviour of aa_splitn_fqname() is expected and checked in
other places where it is called (e.g. aa_remove_profiles). Well, there
is an explicit comment "a ns name without a following profile is allowed"
inside.
AFAICS, nothing can prevent unpacked "name" to be in form like
":samba-dcerpcd" - it is passed from userspace.
Deny the whole profile set replacement in such case and inform user with
EPROTO and an explaining message.
Found by Linux Verification Center (linuxtesting.org).
Fixes: 04dc715e24 ("apparmor: audit policy ns specified in policy load")
Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
Signed-off-by: John Johansen <john.johansen@canonical.com>
If we fail to unpack the transition table then the table elements which
have been already allocated are not freed on error path.
unreferenced object 0xffff88802539e000 (size 128):
comm "apparmor_parser", pid 903, jiffies 4294914938 (age 35.085s)
hex dump (first 32 bytes):
20 73 6f 6d 65 20 6e 61 73 74 79 20 73 74 72 69 some nasty stri
6e 67 20 73 6f 6d 65 20 6e 61 73 74 79 20 73 74 ng some nasty st
backtrace:
[<ffffffff81ddb312>] __kmem_cache_alloc_node+0x1e2/0x2d0
[<ffffffff81c47194>] __kmalloc_node_track_caller+0x54/0x170
[<ffffffff81c225b9>] kmemdup+0x29/0x60
[<ffffffff83e1ee65>] aa_unpack_strdup+0xe5/0x1b0
[<ffffffff83e20808>] unpack_pdb+0xeb8/0x2700
[<ffffffff83e23567>] unpack_profile+0x1507/0x4a30
[<ffffffff83e27bfa>] aa_unpack+0x36a/0x1560
[<ffffffff83e194c3>] aa_replace_profiles+0x213/0x33c0
[<ffffffff83de9461>] policy_update+0x261/0x370
[<ffffffff83de978e>] profile_replace+0x20e/0x2a0
[<ffffffff81eac8bf>] vfs_write+0x2af/0xe00
[<ffffffff81eaddd6>] ksys_write+0x126/0x250
[<ffffffff88f34fb6>] do_syscall_64+0x46/0xf0
[<ffffffff890000ea>] entry_SYSCALL_64_after_hwframe+0x6e/0x76
Call aa_free_str_table() on error path as was done before the blamed
commit. It implements all necessary checks, frees str_table if it is
available and nullifies the pointers.
Found by Linux Verification Center (linuxtesting.org).
Fixes: a0792e2ced ("apparmor: make transition table unpack generic so it can be reused")
Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
Signed-off-by: John Johansen <john.johansen@canonical.com>
There are only two spots where we modify (add to or remove objects from)
the GPIO device list. Readers should be able to access it concurrently.
Replace the mutex with a read-write semaphore and adjust the locking
operations accordingly.
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
If we wait until the GPIO device's .release() callback gets invoked
before we remove it from the global device list, then we risk that
someone will look it up using gpio_device_find() between where we
dropped the last reference and before .release() is done taking a
reference again to an object that's being released.
The device must be removed when it's being unregistered - just like how
we remove it from the GPIO bus.
Fixes: ff2b135922 ("gpio: make the gpiochip a real device")
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
When running the instruction decoder selftest with LLVM=1 and
CONFIG_PVH=y, there is a series of warnings:
arch/x86/tools/insn_decoder_test: warning: Found an x86 instruction decoder bug, please report this.
arch/x86/tools/insn_decoder_test: warning: ffffffff81000050 ea <unknown>
arch/x86/tools/insn_decoder_test: warning: objdump says 1 bytes, but insn_get_length() says 7
arch/x86/tools/insn_decoder_test: warning: Decoded and checked 7214721 instructions with 1 failures
GNU objdump outputs "(bad)" instead of "<unknown>", which is already
handled in the bad_expr regex, so there is no warning.
$ objdump -d arch/x86/platform/pvh/head.o | grep -E '50:\s+ea'
50: ea (bad)
$ llvm-objdump -d arch/x86/platform/pvh/head.o | grep -E '50:\s+ea'
50: ea <unknown>
Add "<unknown>" to the bad_expr regex to clear up the warning, allowing
the instruction decoder selftest to fully pass with llvm-objdump.
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20231205-objdump_reformat-awk-handle-llvm-objdump-bad_expr-v1-1-b4a74f39396f@kernel.org
ASoC: Fixes for v6.7
I recently got a LibreTech Sapphire board for my CI and while
integrating it found and fixed some issues, including crashes for the
enum validation. There's also a couple of patches adding quirks for
another x86 laptop from Hans and an error handling fix for the Freescale
rpmsg driver.
Andrii Nakryiko says:
====================
Libbpf-side __arg_ctx fallback support
Support __arg_ctx global function argument tag semantics even on older kernels
that don't natively support it through btf_decl_tag("arg:ctx").
Patches #2-#6 are preparatory work to allow to postpone BTF loading into the
kernel until after all the BPF program relocations (including global func
appending to main programs) are done. Patch #4 is perhaps the most important
and establishes pre-created stable placeholder FDs, so that relocations can
embed valid map FDs into ldimm64 instructions.
Once BTF is done after relocation, what's left is to adjust BTF information to
have each main program's copy of each used global subprog to point to its own
adjusted FUNC -> FUNC_PROTO type chain (if they use __arg_ctx) in such a way
as to satisfy type expectations of BPF verifier regarding the PTR_TO_CTX
argument definition. See patch #8 for details.
Patch #8 adds few more __arg_ctx use cases (edge cases like multiple arguments
having __arg_ctx, etc) to test_global_func_ctx_args.c, to make it simple to
validate that this logic indeed works on old kernels. It does. But just to be
100% sure patch #9 adds a test validating that libbpf uploads func_info with
properly modified BTF data.
v2->v3:
- drop renaming patch (Alexei, Eduard);
- use memfd_create() instead of /dev/null for placeholder FD (Eduard);
- add one more test for validating BTF rewrite logic (Eduard);
- fixed wrong -errno usage, reshuffled some BTF rewrite bits (Eduard);
v1->v2:
- do internal functions renaming in patch #1 (Alexei);
- extract cloning of FUNC -> FUNC_PROTO information into separate function
(Alexei);
====================
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Link: https://lore.kernel.org/r/20240104013847.3875810-1-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Add a few extra cases of global funcs with context arguments. This time
rely on "arg:ctx" decl_tag (__arg_ctx macro), but put it next to
"classic" cases where context argument has to be of an exact type that
BPF verifier expects (e.g., bpf_user_pt_regs_t for kprobe/uprobe).
Colocating all these cases separately from other global func args that
rely on arg:xxx decl tags (in verifier_global_subprogs.c) allows for
simpler backwards compatibility testing on old kernels. All the cases in
test_global_func_ctx_args.c are supposed to work on older kernels, which
was manually validated during development.
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240104013847.3875810-9-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Out of all special global func arg tag annotations, __arg_ctx is
practically is the most immediately useful and most critical to have
working across multitude kernel version, if possible. This would allow
end users to write much simpler code if __arg_ctx semantics worked for
older kernels that don't natively understand btf_decl_tag("arg:ctx") in
verifier logic.
Luckily, it is possible to ensure __arg_ctx works on old kernels through
a bit of extra work done by libbpf, at least in a lot of common cases.
To explain the overall idea, we need to go back at how context argument
was supported in global funcs before __arg_ctx support was added. This
was done based on special struct name checks in kernel. E.g., for
BPF_PROG_TYPE_PERF_EVENT the expectation is that argument type `struct
bpf_perf_event_data *` mark that argument as PTR_TO_CTX. This is all
good as long as global function is used from the same BPF program types
only, which is often not the case. If the same subprog has to be called
from, say, kprobe and perf_event program types, there is no single
definition that would satisfy BPF verifier. Subprog will have context
argument either for kprobe (if using bpf_user_pt_regs_t struct name) or
perf_event (with bpf_perf_event_data struct name), but not both.
This limitation was the reason to add btf_decl_tag("arg:ctx"), making
the actual argument type not important, so that user can just define
"generic" signature:
__noinline int global_subprog(void *ctx __arg_ctx) { ... }
I won't belabor how libbpf is implementing subprograms, see a huge
comment next to bpf_object_relocate_calls() function. The idea is that
each main/entry BPF program gets its own copy of global_subprog's code
appended.
This per-program copy of global subprog code *and* associated func_info
.BTF.ext information, pointing to FUNC -> FUNC_PROTO BTF type chain
allows libbpf to simulate __arg_ctx behavior transparently, even if the
kernel doesn't yet support __arg_ctx annotation natively.
The idea is straightforward: each time we append global subprog's code
and func_info information, we adjust its FUNC -> FUNC_PROTO type
information, if necessary (that is, libbpf can detect the presence of
btf_decl_tag("arg:ctx") just like BPF verifier would do it).
The rest is just mechanical and somewhat painful BTF manipulation code.
It's painful because we need to clone FUNC -> FUNC_PROTO, instead of
reusing it, as same FUNC -> FUNC_PROTO chain might be used by another
main BPF program within the same BPF object, so we can't just modify it
in-place (and cloning BTF types within the same struct btf object is
painful due to constant memory invalidation, see comments in code).
Uploaded BPF object's BTF information has to work for all BPF
programs at the same time.
Once we have FUNC -> FUNC_PROTO clones, we make sure that instead of
using some `void *ctx` parameter definition, we have an expected `struct
bpf_perf_event_data *ctx` definition (as far as BPF verifier and kernel
is concerned), which will mark it as context for BPF verifier. Same
global subprog relocated and copied into another main BPF program will
get different type information according to main program's type. It all
works out in the end in a completely transparent way for end user.
Libbpf maintains internal program type -> expected context struct name
mapping internally. Note, not all BPF program types have named context
struct, so this approach won't work for such programs (just like it
didn't before __arg_ctx). So native __arg_ctx is still important to have
in kernel to have generic context support across all BPF program types.
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240104013847.3875810-8-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Move the logic of finding and assigning exception callback indices from
BTF sanitization step to program relocations step, which seems more
logical and will unblock moving BTF loading to after relocation step.
Exception callbacks discovery and assignment has no dependency on BTF
being loaded into the kernel, it only uses BTF information. It does need
to happen before subprogram relocations happen, though. Which is why the
split.
No functional changes.
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240104013847.3875810-6-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Move map creation to later during BPF object loading by pre-creating
stable placeholder FDs (utilizing memfd_create()). Use dup2()
syscall to then atomically make those placeholder FDs point to real
kernel BPF map objects.
This change allows to delay BPF map creation to after all the BPF
program relocations. That, in turn, allows to delay BTF finalization and
loading into kernel to after all the relocations as well. We'll take
advantage of the latter in subsequent patches to allow libbpf to adjust
BTF in a way that helps with BPF global function usage.
Clean up a few places where we close map->fd, which now shouldn't
happen, because map->fd should be a valid FD regardless of whether map
was created or not. Surprisingly and nicely it simplifies a bunch of
error handling code. If this change doesn't backfire, I'm tempted to
pre-create such stable FDs for other entities (progs, maybe even BTF).
We previously did some manipulations to make gen_loader work with fake
map FDs, with stable map FDs this hack is not necessary for maps (we
still have it for BTF, but I left it as is for now).
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240104013847.3875810-5-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
With the upcoming switch to preallocated placeholder FDs for maps,
switch various getters/setter away from checking map->fd. Use
map_is_created() helper that detect whether BPF map can be modified based
on map->obj->loaded state, with special provision for maps set up with
bpf_map__reuse_fd().
For backwards compatibility, we take map_is_created() into account in
bpf_map__fd() getter as well. This way before bpf_object__load() phase
bpf_map__fd() will always return -1, just as before the changes in
subsequent patches adding stable map->fd placeholders.
We also get rid of all internal uses of bpf_map__fd() getter, as it's
more oriented for uses external to libbpf. The above map_is_created()
check actually interferes with some of the internal uses, if map FD is
fetched through bpf_map__fd().
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20240104013847.3875810-4-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Yonghong Song says:
====================
bpf: Reduce memory usage for bpf_global_percpu_ma
Currently when a bpf program intends to allocate memory for percpu kptr,
the verifier will call bpf_mem_alloc_init() to prefill all supported
unit sizes and this caused memory consumption very big for large number
of cpus. For example, for 128-cpu system, the total memory consumption
with initial prefill is ~175MB. Things will become worse for systems
with even more cpus.
Patch 1 avoids unnecessary extra percpu memory allocation.
Patch 2 adds objcg to bpf_mem_alloc at init stage so objcg can be
associated with root cgroup and objcg can be passed to later
bpf_mem_alloc_percpu_unit_init().
Patch 3 addresses memory consumption issue by avoiding to prefill
with all unit sizes, i.e. only prefilling with user specified size.
Patch 4 further reduces memory consumption by limiting the
number of prefill entries for percpu memory allocation.
Patch 5 has much smaller low/high watermarks for percpu allocation
to reduce memory consumption.
Patch 6 rejects percpu memory allocation with bpf_global_percpu_ma
when allocation size is greater than 512 bytes.
Patch 7 fixed test_bpf_ma test due to Patch 5.
Patch 8 added one test to show the verification failure log message.
Changelogs:
v5 -> v6:
. Change bpf_mem_alloc_percpu_init() to add objcg as one of parameters.
For bpf_global_percpu_ma, the objcg is NULL, corresponding root memcg.
v4 -> v5:
. Do not do bpf_global_percpu_ma initialization at init stage, instead
doing initialization when the verifier knows it is going to be used
by bpf prog.
. Using much smaller low/high watermarks for percpu allocation.
v3 -> v4:
. Add objcg to bpf_mem_alloc during init stage.
. Initialize objcg at init stage but use it in bpf_mem_alloc_percpu_unit_init().
. Remove check_obj_size() in bpf_mem_alloc_percpu_unit_init().
v2 -> v3:
. Clear the bpf_mem_cache if prefill fails.
. Change test_bpf_ma percpu allocation tests to use bucket_size
as allocation size instead of bucket_size - 8.
. Remove __GFP_ZERO flag from __alloc_percpu_gfp() call.
v1 -> v2:
. Avoid unnecessary extra percpu memory allocation.
. Add a separate function to do bpf_global_percpu_ma initialization
. promote.
. Promote function static 'sizes' array to file static.
. Add comments to explain to refill only one item for percpu alloc.
====================
Link: https://lore.kernel.org/r/20231222031729.1287957-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
In the previous patch, the maximum data size for bpf_global_percpu_ma
is 512 bytes. This breaks selftest test_bpf_ma. The test is adjusted
in two aspects:
- Since the maximum allowed data size for bpf_global_percpu_ma is
512, remove all tests beyond that, names sizes 1024, 2048 and 4096.
- Previously the percpu data size is bucket_size - 8 in order to
avoid percpu allocation into the next bucket. This patch removed
such data size adjustment thanks to Patch 1.
Also, a better way to generate BTF type is used than adding
a member to the value struct.
Acked-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20231222031807.1292853-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
For percpu data structure allocation with bpf_global_percpu_ma,
the maximum data size is 4K. But for a system with large
number of cpus, bigger data size (e.g., 2K, 4K) might consume
a lot of memory. For example, the percpu memory consumption
with unit size 2K and 1024 cpus will be 2K * 1K * 1k = 2GB
memory.
We should discourage such usage. Let us limit the maximum data
size to be 512 for bpf_global_percpu_ma allocation.
Acked-by: Hou Tao <houtao1@huawei.com>
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Link: https://lore.kernel.org/r/20231222031801.1290841-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Currently, refill low/high marks are set with the assumption
of normal non-percpu memory allocation. For example, for
an allocation size 256, for non-percpu memory allocation,
low mark is 32 and high mark is 96, resulting in the
batch allocation of 48 elements and the allocated memory
will be 48 * 256 = 12KB for this particular cpu.
Assuming an 128-cpu system, the total memory consumption
across all cpus will be 12K * 128 = 1.5MB memory.
This might be okay for non-percpu allocation, but may not be
good for percpu allocation, which will consume 1.5MB * 128 = 192MB
memory in the worst case if every cpu has a chance of memory
allocation.
In practice, percpu allocation is very rare compared to
non-percpu allocation. So let us have smaller low/high marks
which can avoid unnecessary memory consumption.
Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Acked-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20231222031755.1289671-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Rafael J. Wysocki
Vlastimil Babka
Russell King (Oracle)
Michael Chan
Claudiu Beznea
Chen Ni
David S. Miller
Jiawen Wu
Naveen Mamindlapalli
Eric Dumazet
Fedor Pchelkin
Bartosz Golaszewski
Nathan Chancellor
Siddhesh Dharme
Takashi Iwai
Alexei Starovoitov
Andrii Nakryiko
Jinghao Jia
Yonghong Song