| Age | Commit message (Collapse) | Author |
|---|
|
Recent kernels cause a lot of TCP retransmissions
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 2.24 GBytes 19.2 Gbits/sec 2767 442 KBytes
[ 5] 1.00-2.00 sec 2.23 GBytes 19.1 Gbits/sec 2312 350 KBytes
^^^^
Replacing the qdisc with pfifo makes retransmissions go away.
It appears that a flow may have a delayed packet with a very near
Tx time. Later, we may get busy processing Rx and the target Tx time
will pass, but we won't service Tx since the CPU is busy with Rx.
If Rx sees an ACK and we try to push more data for the delayed flow
we may fastpath the skb, not realizing that there are already "ready
to send" packets for this flow sitting in the qdisc.
Don't trust the fastpath if we are "behind" according to the projected
Tx time for next flow waiting in the Qdisc. Because we consider anything
within the offload window to be okay for fastpath we must consider
the entire offload window as "now".
Qdisc config:
qdisc fq 8001: dev eth0 parent 1234:1 limit 10000p flow_limit 100p \
buckets 32768 orphan_mask 1023 bands 3 \
priomap 1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1 \
weights 589824 196608 65536 quantum 3028b initial_quantum 15140b \
low_rate_threshold 550Kbit \
refill_delay 40ms timer_slack 10us horizon 10s horizon_drop
For iperf this change seems to do fine, the reordering is gone.
The fastpath still gets used most of the time:
gc 0 highprio 0 fastpath 142614 throttled 418309 latency 19.1us
xx_behind 2731
where "xx_behind" counts how many times we hit the new "return false".
CC: stable@vger.kernel.org
Fixes: 076433bd78d7 ("net_sched: sch_fq: add fast path for mostly idle qdisc")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20241124022148.3126719-1-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
The cited commit replaced inet_csk_reqsk_queue_drop_and_put() with
__inet_csk_reqsk_queue_drop() and reqsk_put() in reqsk_timer_handler().
Then, oreq should be passed to reqsk_put() instead of req; otherwise
use-after-free of nreq could happen when reqsk is migrated but the
retry attempt failed (e.g. due to timeout).
Let's pass oreq to reqsk_put().
Fixes: e8c526f2bdf1 ("tcp/dccp: Don't use timer_pending() in reqsk_queue_unlink().")
Reported-by: Liu Jian <liujian56@huawei.com>
Closes: https://lore.kernel.org/netdev/1284490f-9525-42ee-b7b8-ccadf6606f6d@huawei.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Reviewed-by: Liu Jian <liujian56@huawei.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://patch.msgid.link/20241123174236.62438-1-kuniyu@amazon.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
When phy_ethtool_set_eee_noneg() detects a change in the LPI
parameters, it attempts to update phylib state and trigger the link
to cycle so the MAC sees the updated parameters.
However, in doing so, it sets phydev->enable_tx_lpi depending on
whether the EEE configuration allows the MAC to generate LPI without
taking into account the result of negotiation.
This can be demonstrated with a 1000base-T FD interface by:
# ethtool --set-eee eno0 advertise 8 # cause EEE to be not negotiated
# ethtool --set-eee eno0 tx-lpi off
# ethtool --set-eee eno0 tx-lpi on
This results in being true, despite EEE not having been negotiated and:
# ethtool --show-eee eno0
EEE status: enabled - inactive
Tx LPI: 250 (us)
Supported EEE link modes: 100baseT/Full
1000baseT/Full
Advertised EEE link modes: 100baseT/Full
1000baseT/Full
Fix this by keeping track of whether EEE was negotiated via a new
eee_active member in struct phy_device, and include this state in
the decision whether phydev->enable_tx_lpi should be set.
Fixes: 3e43b903da04 ("net: phy: Immediately call adjust_link if only tx_lpi_enabled changes")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/E1tErSe-005RhB-2R@rmk-PC.armlinux.org.uk
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth
Luiz Augusto von Dentz says:
====================
bluetooth pull request for net:
- SCO: remove the redundant sco_conn_put
- MGMT: Fix slab-use-after-free Read in set_powered_sync
- MGMT: Fix possible deadlocks
* tag 'for-net-2024-11-26' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth:
Bluetooth: SCO: remove the redundant sco_conn_put
Bluetooth: MGMT: Fix possible deadlocks
Bluetooth: MGMT: Fix slab-use-after-free Read in set_powered_sync
====================
Link: https://patch.msgid.link/20241126165149.899213-1-luiz.dentz@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Michal Luczaj says:
====================
net: Fix some callers of copy_from_sockptr()
Some callers misinterpret copy_from_sockptr()'s return value. The function
follows copy_from_user(), i.e. returns 0 for success, or the number of
bytes not copied on error. Simply returning the result in a non-zero case
isn't usually what was intended.
Compile tested with CONFIG_LLC, CONFIG_AF_RXRPC, CONFIG_BT enabled.
Last patch probably belongs more to net-next, if any. Here as an RFC.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Michal Luczaj <mhal@rbox.co>
====================
Link: https://patch.msgid.link/20241119-sockptr-copy-fixes-v3-0-d752cac4be8e@rbox.co
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
copy_from_sockptr() has a history of misuse. Add a comment explaining that
the function follows API of copy_from_user(), i.e. returns 0 for success,
or number of bytes not copied on error.
Signed-off-by: Michal Luczaj <mhal@rbox.co>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
copy_from_sockptr() does not return negative value on error; instead, it
reports the number of bytes that failed to copy. Since it's deprecated,
switch to copy_safe_from_sockptr().
Note: Keeping the `optlen != sizeof(unsigned int)` check as
copy_safe_from_sockptr() by itself would also accept
optlen > sizeof(unsigned int). Which would allow a more lenient handling
of inputs.
Fixes: 17926a79320a ("[AF_RXRPC]: Provide secure RxRPC sockets for use by userspace and kernel both")
Signed-off-by: Michal Luczaj <mhal@rbox.co>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
copy_from_sockptr() is used incorrectly: return value is the number of
bytes that could not be copied. Since it's deprecated, switch to
copy_safe_from_sockptr().
Note: Keeping the `optlen != sizeof(int)` check as copy_safe_from_sockptr()
by itself would also accept optlen > sizeof(int). Which would allow a more
lenient handling of inputs.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Suggested-by: David Wei <dw@davidwei.uk>
Signed-off-by: Michal Luczaj <mhal@rbox.co>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
|
|
Medion E15433 laptop wich ALC269VC (SSID 2782:1705) needs the same
workaround for the missing speaker as another model.
Link: https://bugzilla.suse.com/show_bug.cgi?id=1233298
Cc: <stable@vger.kernel.org>
Link: https://patch.msgid.link/20241128072646.15659-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
|
|
Change the naming for consistency.
While at this, fix the comments in scripts/link-vmlinux.sh.
Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
W=1 builds reported warnings regarding files being ignored:
tools/testing/selftests/arm64/tags/.gitignore: warning: ignored by one of the .gitignore files
tools/testing/selftests/arm64/tags/Makefile: warning: ignored by one of the .gitignore files
tools/testing/selftests/arm64/tags/tags_test.c: warning: ignored by one of the .gitignore files
Adjusting the .gitignore entries will prevent these warnings and
ensure a smoother script execution.
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
resolve_btfids is used by link-vmlinux.sh.
In contrast to other configuration options and targets no transitive
dependency between resolve_btfids and vmlinux.
Add an explicit one.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Use a helper available in scripts/include/hash.h.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Python3 is necessary for running some scripts such as
drivers/gpu/drm/msm/registers/gen_header.py
Both scripts/package/kernel.spec and scripts/package/PKGBUILD already
list Python as the build dependency.
Do likewise for scripts/package/mkdebian.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
Modify this function to return earlier when find_symbol() returns NULL,
reducing the level of improve readability.
No functional changes are intended.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
The first error message in device_id_check() is obscure and can be
misleading because the cause of the error is unlikely to be found in
the struct definition in mod_devicetable.h.
This type of error occurs when an array is passed to an incorrect type
of MODULE_DEVICE_TABLE().
[Example 1]
static const struct acpi_device_id foo_ids[] = {
{ "FOO" },
{ /* sentinel */ },
};
MODULE_DEVICE_TABLE(of, foo_ids);
Currently, modpost outputs a meaningless suggestion:
ERROR: modpost: ...: sizeof(struct of_device_id)=200 is not a modulo of the size of section __mod_device_table__of__<identifier>=64.
Fix definition of struct of_device_id in mod_devicetable.h
The root cause here is that MODULE_DEVICE_TABLE(of, ...) is used instead
of the correct MODULE_DEVICE_TABLE(acpi, ...).
This commit provides a more intuitive error message:
ERROR: modpost: ...: type mismatch between foo_ids[] and MODULE_DEVICE_TABLE(of, ...)
The second error message, related to a missing terminator, is too
verbose.
[Example 2]
static const struct acpi_device_id foo_ids[] = {
{ "FOO" },
};
MODULE_DEVICE_TABLE(acpi, foo_ids);
The current error message is overly long, and does not pinpoint the
incorrect array:
...: struct acpi_device_id is 32 bytes. The last of 1 is:
0x46 0x4f 0x4f 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
ERROR: modpost: ...: struct acpi_device_id is not terminated with a NULL entry!
This commit changes it to a more concise error message, sufficient to
identify the incorrect array:
ERROR: modpost: ...: foo_ids[] is not terminated with a NULL entry
Lastly, this commit squashes device_id_check() into do_table() and
changes fatal() into error(), allowing modpost to continue processing
other modules.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
This commit renames the alias symbol, __mod_<type>__<name>_device_table
to __mod_device_table__<type>__<name>.
This change simplifies the code slightly, as there is no longer a need
to check both the prefix and suffix.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
This commit renames the variables in handle_moddevtable() as follows:
name -> type
namelen -> typelen
identifier -> name
These changes align with the definition in include/linux/module.h:
extern typeof(name) __mod_##type##__##name##_device_table
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
This macro is useful in file2alias.c as well.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
do_usb_table() no longer needs to iterate over the usb_device_id array.
Convert it to a generic ->do_entry() handler.
This is the last special case. Clean up handle_moddevtable().
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
do_of_table() no longer needs to iterate over the of_device_id array.
Convert it to a generic ->do_entry() handler.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
do_pnp_device_entry() no longer needs to iterate over the
pnp_device_id array.
Convert it to a generic ->do_entry() handler.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
do_pnp_card_entries() no longer needs to iterate over the
pnp_card_device_id array.
Convert it to a generic ->do_entry() handler.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
The do_*_entry() functions cannot check the length of the given buffer.
Use module_alias_printf() helper consistently for these functions.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Replace the first argument with a pointer to struct module.
'filename' can be replaced with mod->name.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
With the former cleanups in do_pnp_card_entries(), this macro is no
longer used by anyone.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
MODULE_DEVICE_TABLE(pnp_card, ...) may have duplicated IDs. For
instance, snd_ad1816a_pnpids[] in sound/isa/ad1816a/ad1816a.c includes
multiple occurrences of the "ADS7180" string within its .devs fields.
Currently, do_pnp_card_entries() handles deduplication on its own, but
this logic should be moved to a common helper function, as drivers in
other subsystems might also have similar duplication issues.
For example, drivers/media/i2c/s5c73m3/s5c73m3.mod.c contains duplicated
MODULE_ALIAS() entries because both s5c73m3-core.c and s5c73m3-spi.c
define the same compatible string.
This commit eliminates redundant MODULE_ALIAS() entries across all
drivers.
[Before]
$ grep MODULE_ALIAS drivers/media/i2c/s5c73m3/s5c73m3.mod.c
MODULE_ALIAS("i2c:S5C73M3");
MODULE_ALIAS("of:N*T*Csamsung,s5c73m3");
MODULE_ALIAS("of:N*T*Csamsung,s5c73m3C*");
MODULE_ALIAS("of:N*T*Csamsung,s5c73m3");
MODULE_ALIAS("of:N*T*Csamsung,s5c73m3C*");
[After]
$ grep MODULE_ALIAS drivers/media/i2c/s5c73m3/s5c73m3.mod.c
MODULE_ALIAS("i2c:S5C73M3");
MODULE_ALIAS("of:N*T*Csamsung,s5c73m3");
MODULE_ALIAS("of:N*T*Csamsung,s5c73m3C*");
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
The generic ->do_entry() handler is currently limited to returning
a single alias string.
However, this is not flexible enough for several subsystems, which
currently require their own implementations:
- do_usb_table()
- do_of_table()
- do_pnp_device_entry()
- do_pnp_card_entries()
This commit introduces a helper function so that these special cases can
add multiple MODULE_ALIAS() and then migrate to the generic framework.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
The 'id' pointer is never NULL since it has the same address as
'symval'.
Also, checking (*id)[0] is simpler than calling strlen().
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
This function contains multiple bugs after the following commits:
- ac551828993e ("modpost: i2c aliases need no trailing wildcard")
- 6543becf26ff ("mod/file2alias: make modalias generation safe for cross compiling")
Commit ac551828993e inserted the following code to do_eisa_entry():
else
strcat(alias, "*");
This is incorrect because 'alias' is uninitialized. If it is not
NULL-terminated, strcat() could cause a buffer overrun.
Even if 'alias' happens to be zero-filled, it would output:
MODULE_ALIAS("*");
This would match anything. As a result, the module could be loaded by
any unrelated uevent from an unrelated subsystem.
Commit ac551828993e introduced another bug.
Prior to that commit, the conditional check was:
if (eisa->sig[0])
This checked if the first character of eisa_device_id::sig was not '\0'.
However, commit ac551828993e changed it as follows:
if (sig[0])
sig[0] is NOT the first character of the eisa_device_id::sig. The
type of 'sig' is 'char (*)[8]', meaning that the type of 'sig[0]' is
'char [8]' instead of 'char'. 'sig[0]' and 'symval' refer to the same
address, which never becomes NULL.
The correct conversion would have been:
if ((*sig)[0])
However, this if-conditional was meaningless because the earlier change
in commit ac551828993e was incorrect.
This commit removes the entire incorrect code, which should never have
been executed.
Fixes: ac551828993e ("modpost: i2c aliases need no trailing wildcard")
Fixes: 6543becf26ff ("mod/file2alias: make modalias generation safe for cross compiling")
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Set the -e option to ensure this script fails on any unexpected errors.
Without this change, the kernel build may continue running with an
incorrect string in include/config/kernel.release.
Currently, try_tag() returns 1 when the expected tag is not found as an
ancestor, but this is a case where the script should continue.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Contrary to expectations, passing a single candidate tag to "git
describe" is slower than not passing any --match options.
$ time git describe --debug
...
traversed 10619 commits
...
v6.12-rc5-63-g0fc810ae3ae1
real 0m0.169s
$ time git describe --match=v6.12-rc5 --debug
...
traversed 1310024 commits
v6.12-rc5-63-g0fc810ae3ae1
real 0m1.281s
In fact, the --debug output shows that git traverses all or most of
history. For some repositories and/or git versions, those 1.3s are
actually 10-15 seconds.
This has been acknowledged as a performance bug in git [1], and a fix
is on its way [2]. However, no solution is yet in git.git, and even
when one lands, it will take quite a while before it finds its way to
a release and for $random_kernel_developer to pick that up.
So rewrite the logic to use plumbing commands. For each of the
candidate values of $tag, we ask: (1) is $tag even an annotated
tag? (2) Is it eligible to describe HEAD, i.e. an ancestor of
HEAD? (3) If so, how many commits are in $tag..HEAD?
I have tested that this produces the same output as the current script
for ~700 random commits between v6.9..v6.10. For those 700 commits,
and in my git repo, the 'make -s kernelrelease' command is on average
~4 times faster with this patch applied (geometric mean of ratios).
For the commit mentioned in Josh's original report [3], the
time-consuming part of setlocalversion goes from
$ time git describe --match=v6.12-rc5 c1e939a21eb1
v6.12-rc5-44-gc1e939a21eb1
real 0m1.210s
to
$ time git rev-list --count --left-right v6.12-rc5..c1e939a21eb1
0 44
real 0m0.037s
[1] https://lore.kernel.org/git/20241101113910.GA2301440@coredump.intra.peff.net/
[2] https://lore.kernel.org/git/20241106192236.GC880133@coredump.intra.peff.net/
[3] https://lore.kernel.org/lkml/309549cafdcfe50c4fceac3263220cc3d8b109b2.1730337435.git.jpoimboe@kernel.org/
Reported-by: Sean Christopherson <seanjc@google.com>
Closes: https://lore.kernel.org/lkml/ZPtlxmdIJXOe0sEy@google.com/
Reported-by: Josh Poimboeuf <jpoimboe@kernel.org>
Closes: https://lore.kernel.org/lkml/309549cafdcfe50c4fceac3263220cc3d8b109b2.1730337435.git.jpoimboe@kernel.org/
Tested-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Replace lz4c with lz4 for kernel image compression.
Although lz4 and lz4c are functionally similar, lz4c has been deprecated
upstream since 2018. Since as early as Ubuntu 16.04 and Fedora 25, lz4
and lz4c have been packaged together, making it safe to update the
requirement from lz4c to lz4.
Consequently, some distributions and build systems, such as OpenEmbedded,
have fully transitioned to using lz4. OpenEmbedded core adopted this
change in commit fe167e082cbd ("bitbake.conf: require lz4 instead of
lz4c"), causing compatibility issues when building the mainline kernel
in the latest OpenEmbedded environment, as seen in the errors below.
This change also updates the LZ4 compression commands to make it backward
compatible by replacing stdin and stdout with the '-' option, due to some
unclear reason, the stdout keyword does not work for lz4 and '-' works for
both. In addition, this modifies the legacy '-c1' with '-9' which is also
compatible with both. This fixes the mainline kernel build failures with
the latest master OpenEmbedded builds associated with the mentioned
compatibility issues.
LZ4 arch/arm/boot/compressed/piggy_data
/bin/sh: 1: lz4c: not found
...
...
ERROR: oe_runmake failed
Link: https://github.com/lz4/lz4/pull/553
Suggested-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Signed-off-by: Parth Pancholi <parth.pancholi@toradex.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
This reverts commit 54babdc0343f ("kbuild: Disable KCSAN for
autogenerated *.mod.c intermediaries").
Now that objtool is enabled for *.mod.c, there is no need to filter
out CFLAGS_KCSAN.
I no longer see "Unpatched return thunk in use. This should not happen!"
error with KCSAN when loading a module.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
Currently, objtool is disabled in scripts/Makefile.{modfinal,vmlinux}.
This commit moves rule_cc_o_c and rule_as_o_S to scripts/Makefile.lib
and set objtool-enabled to y there.
With this change, *.mod.o, .module-common.o, builtin-dtb.o, and
vmlinux.export.o will now be covered by objtool.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
The cmd_cc_o_c and cmd_as_o_S macros are duplicated in
scripts/Makefile.{build,modfinal,vmlinux}.
This commit factors them out to scripts/Makefile.lib.
No functional changes are intended.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
|
|
This rule is unnecessary because you can generate foo/bar.symtypes
as a side effect using:
$ make KBUILD_SYMTYPES=1 foo/bar.o
While compiling *.o is slower than preprocessing, the impact is
negligible. I prioritize keeping the code simpler.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
There is no need to pass '-r /dev/null', which is no-op.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
Unless an explicit O= option is provided, external module builds must
start from the kernel directory.
This can be achieved by using the -C option:
$ make -C /path/to/kernel M=/path/to/external/module
This commit allows starting external module builds from any directory,
so you can also do the following:
$ make -f /path/to/kernel/Makefile M=/path/to/external/module
The key difference is that the -C option changes the working directory
and parses the Makefile located there, while the -f option only
specifies the Makefile to use.
As shown in the examples in Documentation/kbuild/modules.rst, external
modules usually have a wrapper Makefile that allows you to build them
without specifying any make arguments. The Makefile typically contains
a rule as follows:
KDIR ?= /path/to/kernel
default:
$(MAKE) -C $(KDIR) M=$(CURDIR) $(MAKECMDGOALS)
The log will appear as follows:
$ make
make -C /path/to/kernel M=/path/to/external/module
make[1]: Entering directory '/path/to/kernel'
make[2]: Entering directory '/path/to/external/module'
CC [M] helloworld.o
MODPOST Module.symvers
CC [M] helloworld.mod.o
CC [M] .module-common.o
LD [M] helloworld.ko
make[2]: Leaving directory '/path/to/external/module'
make[1]: Leaving directory '/path/to/kernel'
This changes the working directory twice because the -C option first
switches to the kernel directory, and then Kbuild internally recurses
back to the external module directory.
With this commit, the wrapper Makefile can directly include the kernel
Makefile:
KDIR ?= /path/to/kernel
export KBUILD_EXTMOD := $(realpath $(dir $(lastword $(MAKEFILE_LIST))))
include $(KDIR)/Makefile
This avoids unnecessary sub-make invocations:
$ make
CC [M] helloworld.o
MODPOST Module.symvers
CC [M] helloworld.mod.o
CC [M] .module-common.o
LD [M] helloworld.ko
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
When Kbuild starts building in a separate output directory, it generates
a wrapper Makefile, allowing you to invoke 'make' from the output
directory.
This commit makes it more convenient, so you can invoke 'make' without
M= or MO=.
First, you need to build external modules in a separate directory:
$ make M=/path/to/module/source/dir MO=/path/to/module/build/dir
Once the wrapper Makefile is generated in /path/to/module/build/dir,
you can proceed as follows:
$ cd /path/to/module/build/dir
$ make
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
Keep the consistent behavior when this Makefile is invoked from another
directory.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
This commit makes -fmacro-prefix-map work for external modules built in
a separate output directory. It improves the reproducibility of external
modules and provides the benefits described in commit a73619a845d5
("kbuild: use -fmacro-prefix-map to make __FILE__ a relative path").
When building_out_of_srctree is not defined (e.g., when the kernel or
external module is built in the source directory), this option is
unnecessary.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
There has been a long-standing request to support building external
modules in a separate build directory.
This commit introduces a new environment variable, KBUILD_EXTMOD_OUTPUT,
and its shorthand Make variable, MO.
A simple usage:
$ make -C <kernel-dir> M=<module-src-dir> MO=<module-build-dir>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
With the previous changes, $(extmod_prefix), $(MODORDER), and
$(MODULES_NSDEPS) are constant. (empty, modules.order, and
modules.nsdeps, respectively).
Remove these variables and hard-code their values.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
|
|
Currently, Kbuild always operates in the output directory of the kernel,
even when building external modules. This increases the risk of external
module Makefiles attempting to write to the kernel directory.
This commit switches the working directory to the external module
directory, allowing the removal of the $(KBUILD_EXTMOD)/ prefix from
some build artifacts.
The command for building external modules maintains backward
compatibility, but Makefiles that rely on working in the kernel
directory may break. In such cases, $(objtree) and $(srctree) should
be used to refer to the output and source directories of the kernel.
The appearance of the build log will change as follows:
[Before]
$ make -C /path/to/my/linux M=/path/to/my/externel/module
make: Entering directory '/path/to/my/linux'
CC [M] /path/to/my/externel/module/helloworld.o
MODPOST /path/to/my/externel/module/Module.symvers
CC [M] /path/to/my/externel/module/helloworld.mod.o
CC [M] /path/to/my/externel/module/.module-common.o
LD [M] /path/to/my/externel/module/helloworld.ko
make: Leaving directory '/path/to/my/linux'
[After]
$ make -C /path/to/my/linux M=/path/to/my/externel/module
make: Entering directory '/path/to/my/linux'
make[1]: Entering directory '/path/to/my/externel/module'
CC [M] helloworld.o
MODPOST Module.symvers
CC [M] helloworld.mod.o
CC [M] .module-common.o
LD [M] helloworld.ko
make[1]: Leaving directory '/path/to/my/externel/module'
make: Leaving directory '/path/to/my/linux'
Printing "Entering directory" twice is cumbersome. This will be
addressed later.
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <n.schier@avm.de>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull more ACPI updates from Rafael Wysocki:
"These add a common init function for arch-specific ACPI
initialization, clean up idle states initialization in the ACPI
processor_idle driver and update quirks:
- Introduce acpi_arch_init() for architecture-specific ACPI subsystem
initialization (Miao Wang)
- Clean up Asus quirks in acpi_quirk_skip_dmi_ids[] and add a quirk
to skip I2C clients on Acer Iconia One 8 A1-840 (Hans de Goede)
- Make the ACPI processor_idle driver use acpi_idle_play_dead() for
all idle states regardless of their types (Rafael Wysocki)"
* tag 'acpi-6.13-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
ACPI: introduce acpi_arch_init()
ACPI: x86: Clean up Asus entries in acpi_quirk_skip_dmi_ids[]
ACPI: x86: Add skip i2c clients quirk for Acer Iconia One 8 A1-840
ACPI: processor_idle: Use acpi_idle_play_dead() for all C-states
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull morepower management updates from Rafael Wysocki:
"These update the OPP (Operating Performance Points) DT bindings for
ti-cpu (Dhruva Gole) and remove unused declarations from the OPP
header file (Zhang Zekun)"
* tag 'pm-6.13-rc1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
dt-bindings: opp: operating-points-v2-ti-cpu: Describe opp-supported-hw
OPP: Remove unused declarations in header file
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Pull more thermal control updates from Rafael Wysocki:
"These fix a Power Allocator thermal governor issue reported recently,
update the Intel int3400 thermal driver and simplify DT data parsing
in the thermal control subsystem:
- Add a NULL pointer check that was missed by recent modifications of
the Power Allocator thermal governor (Rafael Wysocki)
- Remove the data_vault attribute_group from int3400 because it is
only used for exposing one binary file that can be exposed directly
(Thomas Weißschuh)
- Prevent the current_uuid sysfs attribute in int3400 from mistakenly
treating valid UUID values as invalid on some older systems
(Srinivas Pandruvada)
- Use the cleanup.h mechanics to simplify DT data parsing in the
thermal core and some drivers (Krzysztof Kozlowski)"
* tag 'thermal-6.13-rc1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
thermal: sun8i: Use scoped device node handling to simplify error paths
thermal: tegra: Simplify with scoped for each OF child loop
thermal: qcom-spmi-adc-tm5: Simplify with scoped for each OF child loop
thermal: of: Use scoped device node handling to simplify of_thermal_zone_find()
thermal: of: Use scoped memory and OF handling to simplify thermal_of_trips_init()
thermal: of: Simplify thermal_of_should_bind with scoped for each OF child
thermal: gov_power_allocator: Add missing NULL pointer check
thermal: int3400: Remove unneeded data_vault attribute_group
thermal: int3400: Fix reading of current_uuid for active policy
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd
Pull more iommufd updates from Jason Gunthorpe:
"Change the driver callback op domain_alloc_user() into two ops:
domain_alloc_paging_flags() and domain_alloc_nesting() that better
describe what the ops are expected to do.
There will be per-driver cleanup based on this going into the next
cycle via the driver trees"
* tag 'for-linus-iommufd' of git://git.kernel.org/pub/scm/linux/kernel/git/jgg/iommufd:
iommu: Rename ops->domain_alloc_user() to domain_alloc_paging_flags()
iommu: Add ops->domain_alloc_nested()
|
|
A write which goes past the end of the bdev in blkdev_write_iter() will
be truncated. Truncating cannot tolerated for an atomic write, so error
that condition.
Fixes: caf336f81b3a ("block: Add fops atomic write support")
Signed-off-by: John Garry <john.g.garry@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20241127092318.632790-1-john.g.garry@oracle.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
