| Message ID | 1448421962-6174-1-git-send-email-ruscur@russell.cc (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
On 25/11/15 14:26, Russell Currey wrote: > On BMC machines, console output is controlled by the OPAL firmware and is > only flushed when its pollers are called. When the kernel is in a panic > state, it no longer calls these pollers and thus console output does not > completely flush, causing some output from the panic to be lost. > > This patch adds a new kmsg_dumper which gets called at panic time to ensure > panic output is not lost. It accomplishes this by calling OPAL_CONSOLE_FLUSH > in the OPAL API, and if that is not available, the pollers are called enough > times to (hopefully) completely flush the buffer. As per our off-list discussions, the commit message should explain that the console is already flushed upon reboot, and thus this patch fixes a problem in the case where the kernel is not set to reboot on panic, or will only reboot after an extended delay. Additionally, you could still miss messages printed in panic() after the dumpers are called, i.e. the "end kernel panic" message. > > Signed-off-by: Russell Currey <ruscur@russell.cc> The code looks reasonable. With the changes I've mentioned above: Reviewed-by: Andrew Donnellan <andrew.donnellan@au1.ibm.com>
Russell Currey <ruscur@russell.cc> writes: > On BMC machines, console output is controlled by the OPAL firmware and is > only flushed when its pollers are called. When the kernel is in a panic > state, it no longer calls these pollers and thus console output does not > completely flush, causing some output from the panic to be lost. > > This patch adds a new kmsg_dumper which gets called at panic time to ensure > panic output is not lost. It accomplishes this by calling OPAL_CONSOLE_FLUSH > in the OPAL API, and if that is not available, the pollers are called enough > times to (hopefully) completely flush the buffer. You may want to add to commit message that this depends on: https://patchwork.ozlabs.org/patch/547379/ (which I'm about to go review) Considering this does address an actual bug, should this go to stable@ ? (Same question for the skiboot patch)
On Fri, 2015-11-27 at 16:40 +1100, Stewart Smith wrote: > Russell Currey <ruscur@russell.cc> writes: > > On BMC machines, console output is controlled by the OPAL firmware and is > > only flushed when its pollers are called. When the kernel is in a panic > > state, it no longer calls these pollers and thus console output does not > > completely flush, causing some output from the panic to be lost. > > > > This patch adds a new kmsg_dumper which gets called at panic time to ensure > > panic output is not lost. It accomplishes this by calling > > OPAL_CONSOLE_FLUSH > > in the OPAL API, and if that is not available, the pollers are called > > enough > > times to (hopefully) completely flush the buffer. > > You may want to add to commit message that this depends on: > https://patchwork.ozlabs.org/patch/547379/ > It doesn't *depend* on it, it's just preferred. It still works with older Skiboot versions. I will mention it in the actual commit message for the next revision, though. > (which I'm about to go review) > > Considering this does address an actual bug, should this go to stable@ ? > > (Same question for the skiboot patch) > Yes, both this patch and the corresponding Skiboot patch should go to stable.
Russell Currey <ruscur@russell.cc> writes: > On Fri, 2015-11-27 at 16:40 +1100, Stewart Smith wrote: >> Russell Currey <ruscur@russell.cc> writes: >> > On BMC machines, console output is controlled by the OPAL firmware and is >> > only flushed when its pollers are called. When the kernel is in a panic >> > state, it no longer calls these pollers and thus console output does not >> > completely flush, causing some output from the panic to be lost. >> > >> > This patch adds a new kmsg_dumper which gets called at panic time to ensure >> > panic output is not lost. It accomplishes this by calling >> > OPAL_CONSOLE_FLUSH >> > in the OPAL API, and if that is not available, the pollers are called >> > enough >> > times to (hopefully) completely flush the buffer. >> >> You may want to add to commit message that this depends on: >> https://patchwork.ozlabs.org/patch/547379/ >> > It doesn't *depend* on it, it's just preferred. It still works with older > Skiboot versions. I will mention it in the actual commit message for the > next revision, though. More meaning that it depends on that patch being merged before, say the PCI hotplug patches, which add a *different* OPAL call at the same ID - and that it depends on *this* patch with *this* API rather than a modified one being merged. Otherwise, if, say, I merged some other new OPAL call into skiboot first, this patch would likely cause all sorts of trouble. It's also a handy hint to Michael to ensure the skiboot side of things is merged and okay before merging kernel things in.
On Fri, 2015-11-27 at 18:21 +1100, Stewart Smith wrote: > Russell Currey <ruscur@russell.cc> writes: > > On Fri, 2015-11-27 at 16:40 +1100, Stewart Smith wrote: > > > Russell Currey <ruscur@russell.cc> writes: > > > > On BMC machines, console output is controlled by the OPAL firmware and > > > > is > > > > only flushed when its pollers are called. When the kernel is in a > > > > panic > > > > state, it no longer calls these pollers and thus console output does > > > > not > > > > completely flush, causing some output from the panic to be lost. > > > > > > > > This patch adds a new kmsg_dumper which gets called at panic time to > > > > ensure > > > > panic output is not lost. It accomplishes this by calling > > > > OPAL_CONSOLE_FLUSH > > > > in the OPAL API, and if that is not available, the pollers are called > > > > enough > > > > times to (hopefully) completely flush the buffer. > > > > > > You may want to add to commit message that this depends on: > > > https://patchwork.ozlabs.org/patch/547379/ > > > > > It doesn't *depend* on it, it's just preferred. It still works with older > > Skiboot versions. I will mention it in the actual commit message for the > > next revision, though. > > More meaning that it depends on that patch being merged before, say the > PCI hotplug patches, which add a *different* OPAL call at the same ID - > and that it depends on *this* patch with *this* API rather than a > modified one being merged. > > Otherwise, if, say, I merged some other new OPAL call into skiboot > first, this patch would likely cause all sorts of trouble. > > It's also a handy hint to Michael to ensure the skiboot side of things > is merged and okay before merging kernel things in. > That's a very good point I hadn't considered. I'll make a note of it next revision, especially given there are some additional changes required on the Skiboot side of things.
diff --git a/arch/powerpc/include/asm/opal-api.h b/arch/powerpc/include/asm/opal-api.h index 8374afe..f8faaae 100644 --- a/arch/powerpc/include/asm/opal-api.h +++ b/arch/powerpc/include/asm/opal-api.h @@ -157,7 +157,8 @@ #define OPAL_LEDS_GET_INDICATOR 114 #define OPAL_LEDS_SET_INDICATOR 115 #define OPAL_CEC_REBOOT2 116 -#define OPAL_LAST 116 +#define OPAL_CONSOLE_FLUSH 117 +#define OPAL_LAST 117 /* Device tree flags */ diff --git a/arch/powerpc/include/asm/opal.h b/arch/powerpc/include/asm/opal.h index 8001159..a5fd407 100644 --- a/arch/powerpc/include/asm/opal.h +++ b/arch/powerpc/include/asm/opal.h @@ -35,6 +35,7 @@ int64_t opal_console_read(int64_t term_number, __be64 *length, uint8_t *buffer); int64_t opal_console_write_buffer_space(int64_t term_number, __be64 *length); +void opal_console_flush(void); int64_t opal_rtc_read(__be32 *year_month_day, __be64 *hour_minute_second_millisecond); int64_t opal_rtc_write(uint32_t year_month_day, @@ -262,6 +263,8 @@ extern int opal_resync_timebase(void); extern void opal_lpc_init(void); +extern void opal_kmsg_init(void); + extern int opal_event_request(unsigned int opal_event_nr); struct opal_sg_list *opal_vmalloc_to_sg_list(void *vmalloc_addr, diff --git a/arch/powerpc/platforms/powernv/Makefile b/arch/powerpc/platforms/powernv/Makefile index 1c8cdb6..b9de7ef 100644 --- a/arch/powerpc/platforms/powernv/Makefile +++ b/arch/powerpc/platforms/powernv/Makefile @@ -2,6 +2,7 @@ obj-y += setup.o opal-wrappers.o opal.o opal-async.o idle.o obj-y += opal-rtc.o opal-nvram.o opal-lpc.o opal-flash.o obj-y += rng.o opal-elog.o opal-dump.o opal-sysparam.o opal-sensor.o obj-y += opal-msglog.o opal-hmi.o opal-power.o opal-irqchip.o +obj-y += opal-kmsg.o obj-$(CONFIG_SMP) += smp.o subcore.o subcore-asm.o obj-$(CONFIG_PCI) += pci.o pci-p5ioc2.o pci-ioda.o diff --git a/arch/powerpc/platforms/powernv/opal-kmsg.c b/arch/powerpc/platforms/powernv/opal-kmsg.c new file mode 100644 index 0000000..bd3b2ee --- /dev/null +++ b/arch/powerpc/platforms/powernv/opal-kmsg.c @@ -0,0 +1,68 @@ +/* + * kmsg dumper that ensures the OPAL console fully flushes panic messages + * + * Author: Russell Currey <ruscur@russell.cc> + * + * Copyright 2015 IBM Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the + * Free Software Foundation; either version 2 of the License, or (at your + * option) any later version. + */ + +#include <linux/kmsg_dump.h> + +#include <asm/opal.h> +#include <asm/opal-api.h> + +/* + * Console output is controlled by OPAL firmware. The kernel regularly calls + * OPAL_POLL_EVENTS, which flushes some console output. In a panic state, + * however, the kernel no longer calls OPAL_POLL_EVENTS and the panic message + * may not be completely printed. This function does not actually dump the + * message, it just ensures that OPAL completely flushes the console buffer. + */ +static void force_opal_console_flush(struct kmsg_dumper *dumper, + enum kmsg_dump_reason reason) +{ + int i; + + /* + * Outside of a panic context the pollers will continue to run, + * so we don't need to do any special flushing. + */ + if (reason != KMSG_DUMP_PANIC) + return; + + if (opal_check_token(OPAL_CONSOLE_FLUSH)) { + opal_console_flush(); + } else { + /* + * If OPAL_CONSOLE_FLUSH is not implemented in the firmware, + * the console can still be flushed by calling the polling + * function enough times to flush the buffer. We don't know + * how much output still needs to be flushed, but we can be + * generous since the kernel is in panic and doesn't need + * to do much else. + */ + printk(KERN_NOTICE "opal: OPAL_CONSOLE_FLUSH missing.\n"); + for (i = 0; i < 1024; i++) { + opal_poll_events(NULL); + } + } +} + +static struct kmsg_dumper opal_kmsg_dumper = { + .dump = force_opal_console_flush +}; + +void __init opal_kmsg_init(void) +{ + int rc; + + /* Add our dumper to the list */ + rc = kmsg_dump_register(&opal_kmsg_dumper); + if (rc != 0) + pr_err("opal: kmsg_dump_register failed; returned %d\n", rc); +} diff --git a/arch/powerpc/platforms/powernv/opal-wrappers.S b/arch/powerpc/platforms/powernv/opal-wrappers.S index b7a464f..e45b88a 100644 --- a/arch/powerpc/platforms/powernv/opal-wrappers.S +++ b/arch/powerpc/platforms/powernv/opal-wrappers.S @@ -301,3 +301,4 @@ OPAL_CALL(opal_flash_erase, OPAL_FLASH_ERASE); OPAL_CALL(opal_prd_msg, OPAL_PRD_MSG); OPAL_CALL(opal_leds_get_ind, OPAL_LEDS_GET_INDICATOR); OPAL_CALL(opal_leds_set_ind, OPAL_LEDS_SET_INDICATOR); +OPAL_CALL(opal_console_flush, OPAL_CONSOLE_FLUSH); diff --git a/arch/powerpc/platforms/powernv/opal.c b/arch/powerpc/platforms/powernv/opal.c index 4296d55..a94a85c 100644 --- a/arch/powerpc/platforms/powernv/opal.c +++ b/arch/powerpc/platforms/powernv/opal.c @@ -758,6 +758,9 @@ static int __init opal_init(void) opal_pdev_init(opal_node, "ibm,opal-flash"); opal_pdev_init(opal_node, "ibm,opal-prd"); + /* Initialise OPAL kmsg dumper for flushing console on panic */ + opal_kmsg_init(); + return 0; } machine_subsys_initcall(powernv, opal_init);
On BMC machines, console output is controlled by the OPAL firmware and is only flushed when its pollers are called. When the kernel is in a panic state, it no longer calls these pollers and thus console output does not completely flush, causing some output from the panic to be lost. This patch adds a new kmsg_dumper which gets called at panic time to ensure panic output is not lost. It accomplishes this by calling OPAL_CONSOLE_FLUSH in the OPAL API, and if that is not available, the pollers are called enough times to (hopefully) completely flush the buffer. Signed-off-by: Russell Currey <ruscur@russell.cc> --- V2: Fix some code style issues, no functional change A patch to add the preferred OPAL call for flushing the console buffer, OPAL_CONSOLE_FLUSH, was recently sent upstream to Skiboot. You can track its progress here: https://patchwork.ozlabs.org/patch/547379/ --- arch/powerpc/include/asm/opal-api.h | 3 +- arch/powerpc/include/asm/opal.h | 3 ++ arch/powerpc/platforms/powernv/Makefile | 1 + arch/powerpc/platforms/powernv/opal-kmsg.c | 68 ++++++++++++++++++++++++++ arch/powerpc/platforms/powernv/opal-wrappers.S | 1 + arch/powerpc/platforms/powernv/opal.c | 3 ++ 6 files changed, 78 insertions(+), 1 deletion(-) create mode 100644 arch/powerpc/platforms/powernv/opal-kmsg.c