Hi,
[email protected] writes:
> The mm-of-the-moment snapshot 2008-08-06-14-20 has been uploaded to
>
> http://userweb.kernel.org/~akpm/mmotm/
>
> It contains the following patches against 2.6.27-rc2:
>
> feature-removal-scheduletxt-remove-the-ncr53c9x-entry.patch
> hugetlb-call-arch_prepare_hugepage-for-surplus-pages.patch
> quota-documentation-for-sending-below-quota-messages-via-netlink-and-tiny-doc-update.patch
> page-allocator-use-no-panic-variant-of-alloc_bootmem-in-alloc_large_system_hash.patch
> cpu-hotplug-s390-doesnt-support-additional_cpus-anymore.patch
> radeonfb-fix-copyarea-for-r300-and-later.patch
> memcg-fix-oops-in-mem_cgroup_shrink_usage.patch
> sm501-update-debugging-low-information-messages.patch
> seq_file-fix-bug-when-seq_read-reads-nothing.patch
> repeatable-slab-corruption-with-ltp-msgctl08.patch
> linux-next.patch
> next-remove-localversion.patch
> acpi-add-checking-for-null-early-param.patch
> mm-only-enforce-acpi-resource-conflict-checks.patch
> calgary-fix-a-comparison-warning-the-pci-calgary-64-driver.patch
> x86-remove-init_mm-export-as-planned-for-2626.patch
> use-warn-in-arch-x86-mm-ioremapc.patch
> use-warn-in-arch-x86-mm-pageattrc.patch
> use-warn-in-arch-x86-kernel.patch
> arch-x86-pci-irqc-attempt-to-clean-up-code-layout.patch
> i386-vmalloc-size-fix.patch
> zero-based-percpu-use-vmlinux_symbol-in-include-asm-generic-vmlinuxldsh.patch
> x86-64-fix-overlap-of-modules-and-fixmap-areas.patch
> x86_64-remove-empty-lines-from-stack-traces-oopses.patch
> x86-tracehook_signal_handler.patch
> x86-tracehook-syscall.patch
> x86-tracehook-asm-syscallh.patch
> x86-signals-use-asm-syscallh.patch
> x86-tracehook-tif_notify_resume.patch
> x86-tracehook-config_have_arch_tracehook.patch
> hda_intel-enable-snoop-for-nvidia-hda-controller.patch
> cifs-remove-global_extern-macro.patch
> cifs-fix-range-check.patch
> pcmcia-pccard-deadlock-fix.patch
> ppc-use-the-common-ascii-hex-helpers.patch
> powerpc-replace-__function__-with-__func__.patch
> v4l-link-tuner-before-saa7134.patch
> v4l-drx397xdc-sparse-annotations.patch
> i2c-renesas-highlander-fpga-smbus-support.patch
> i2c-renesas-highlander-fpga-smbus-support-update.patch
> i2c-renesas-highlander-fpga-smbus-support-update-fix.patch
> hid-wellspring-device-quirks.patch
> add-time_is_after_jiffies-and-others-which-compare-with-jiffies.patch
> ntp-let-update_persistent_clock-sleep.patch
> migrate_timers-add-comment-use-spinlock_irq.patch
> tick-schedc-suppress-needless-timer-reprogramming.patch
> timekeeping-fix-rounding-problem-during-clock-update.patch
> git-ia64.patch
> drivers-input-serio-xilinx_ps2c-fix-warning.patch
> wistron_btns-add-support-for-fujitsu-siemens-amilo-pro-edition-v3505.patch
> elantech-touchpad-driver.patch
> input-touchscreen-driver-add-support-ad7877-touchscreen-driver.patch
> input-bcm5974-055-smoother-motion-irq-simplification.patch
> input-ads7846c-sparse-lock-annotation.patch
> git-jg-misc.patch
> git-kbuild-fixes.patch
> genksyms-track-symbol-checksum-changes.patch
> genksyms-allow-to-ignore-symbol-checksum-changes.patch
> lkdtm-fix-for-config_scsi=n.patch
> git-lblnet.patch
> git-libata-all.patch
> libata-scsi-dont-start-hotplug-work-queue-if-hotplug-is-disabled.patch
> sata_viac-add-support-for-vt8251-fix-the-internal-chips-issue-and.patch
> m32r-remove-the-unused-nohighmem-option.patch
> m32r-dont-offer-config_isa.patch
> mmc-fix-clock-off-in-pxa255-270.patch
> git-ubi.patch
> git-ubi-git-rejects.patch
> configure-out-igmp-support.patch
> acpi-compal-laptop-use-rfkill-switch-subsystem.patch
> forcedeth-power-down-phy-when-interface-is-down.patch
> forcedeth-fix-mac-address-detection-on-network-card-regression-in-2623.patch
> sundance-set-carrier-status-on-link-change-events.patch
> via-velocity-give-a-structure-to-the-rx-tx-fields.patch
> via-velocity-fix-sleep-with-spinlock-bug-during-mtu-change.patch
> hamradio-add-missing-sanity-check-to-tty-operation.patch
> skge-adapt-skge-to-use-reworked-pci-pm.patch
> cassini-use-request_firmware.patch
> pegasus-add-blacklist-support-to-fix-belkin-bluetooth-dongle.patch
> drivers-net-replace-__function__-with-__func__.patch
> e1000e-convert-ndev_-printks-to-something-smaller.patch
> e1000e-perform-basic-82573-eeprom-checks-for-known-issues.patch
> e1000e-fix-drv-load-issues.patch
> e1000e-remove-inapplicable-test-for-ioport.patch
> e1000e-remove-inapplicable-test-for-ioport-checkpatch-fixes.patch
> sky2-adapt-to-use-reworked-pci-pm-code.patch
> e1000e-make-e1000e-default-to-the-same-kconfig-setting-as-e1000.patch
> 3x59x-fix-pci-resource-management.patch
> smc-ultra-get-rid-of-eth%d-message.patch
> nfs-err_ptr-is-expected-on-failure-from-nfs_do_clone_mount.patch
> sunrpc-do-not-pin-sunrpc-module-in-the-memory.patch
> git-parisc.patch
> pciidsh-remove-a-duplicated-symbol.patch
> use-warn-in-kernel-lockdepc.patch
> sched-do_wait_for_common-use-signal_pending_state.patch
> wait_task_inactive-dont-consider-task-nivcsw.patch
> wait_task_inactive-improve-the-returned-value-for-nvcsw-==-0.patch
> wait_task_inactive-fix-the-config_smp-version.patch
> sched-type-fix.patch
> lockdep-fix-combinatorial-explosion-in-lock-subgraph-traversal.patch
> lockdep-fix-combinatorial-explosion-in-lock-subgraph-traversal-fix.patch
> lockdep-lock_set_subclass-reset-a-held-locks-subclass.patch
> lockdep-re-annotate-scheduler-runqueues.patch
> lockdep-shrink-held_lock-structure.patch
> lockdep-shrink-held_lock-structure-fix.patch
> lockdep-shrink-held_lock-structure-fix-2.patch
> lockdep-map_acquire.patch
> lockdep-lock-protection-locks.patch
> lockdep-spin_lock_nest_lock.patch
> netfilter-conntrack_helper-needs-to-include-rculisth.patch
> rcu-just-rename-call_rcu_bh-instead-of-making-it-a-macro.patch
> rcu-trace-fix-possible-mem-leak.patch
> ftrace-new-continue-entry-separate-out-from-trace_entry.patch
> ftrace-printk-formatting-infrastructure.patch
---
The attribute on the function definition breaks compilation. Instead of
removing it, we could use a declaration and add the attribute there, but
then, when someone adds new ftrace_printk()s, she will probably compile
with CONFIG_TRACING and the callsite is garuanteed to be verified during
development.
Signed-off-by: Johannes Weiner <[email protected]>
---
include/linux/ftrace.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -165,7 +165,7 @@ __ftrace_printk(unsigned long ip, const
static inline void
ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
static inline int
-ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 0)))
+ftrace_printk(const char *fmt, ...)
{
return 0;
}
Hi Andrew,
[sorry, screwed up the recpt list before....]
Johannes Weiner <[email protected]> writes:
> Hi,
>
> [email protected] writes:
>
>> ftrace-new-continue-entry-separate-out-from-trace_entry.patch
>> ftrace-printk-formatting-infrastructure.patch
>
> ---
> The attribute on the function definition breaks compilation. Instead of
> removing it, we could use a declaration and add the attribute there, but
> then, when someone adds new ftrace_printk()s, she will probably compile
> with CONFIG_TRACING and the callsite is garuanteed to be verified during
> development.
>
> Signed-off-by: Johannes Weiner <[email protected]>
> ---
> include/linux/ftrace.h | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> --- a/include/linux/ftrace.h
> +++ b/include/linux/ftrace.h
> @@ -165,7 +165,7 @@ __ftrace_printk(unsigned long ip, const
> static inline void
> ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) { }
> static inline int
> -ftrace_printk(const char *fmt, ...) __attribute__ ((format (printf, 1, 0)))
> +ftrace_printk(const char *fmt, ...)
> {
> return 0;
> }
Still needed, get this with mmotm-2008-08-08-00-56 and !CONFIG_TRACING:
In file included from kernel/sched.c:74:
include/linux/ftrace.h:169: error: expected ‘,’ or ‘;’ before ‘{’ token
make[1]: *** [kernel/sched.o] Error 1
make: *** [kernel] Error 2
Hannes
On Mon, 11 Aug 2008, Andrew Morton wrote:
> On Fri, 08 Aug 2008 12:47:52 +0200
> Johannes Weiner <[email protected]> wrote:
> My attempts to generate a suitable config were not happy ones. Feast
> your eyes upon this monstrosity:
>
> Symbol: TRACING [=y]
> Selects: DEBUG_FS && STACKTRACE
> Selected by: FTRACE && HAVE_FTRACE || IRQSOFF_TRACER && TRACE_IRQFLAGS_SUPPORT && GENERIC_TIME && HAVE_FTRACE || PREEMPT_TRACER && GENERIC_TIME && PREEMPT && HAVE_FTRACE || SYSPROF_TRACER && X86 || SCHED_TRACER && HAVE_FTRACE || CONTEXT_SWITCH_TRACER && HAVE_FTRACE || MMIOTRACE && DEBUG_KERNEL && PCI
>
This isn't as bad as it looks.
Yes tracing selects DEBUG_FS and STACKTRACE, but the TRACING is required
to transcend into the kernel/trace directory. That is where the other
tracers live, and some tracers require other options to be set.
ie. FTRACE requires HAVE_FTRACE, PREEMPT_TRACER requires PREEMPT, etc.
I don't really see a problem with those config dependencies.
-- Steve