- Mar 28, 2017
-
-
Borislav Petkov authored
Introduce a simple data structure for collecting correctable errors along with accessors. More detailed description in the code itself. The error decoding is done with the decoding chain now and mce_first_notifier() gets to see the error first and the CEC decides whether to log it and then the rest of the chain doesn't hear about it - basically the main reason for the CE collector - or to continue running the notifiers. When the CEC hits the action threshold, it will try to soft-offine the page containing the ECC and then the whole decoding chain gets to see the error. Signed-off-by:
Borislav Petkov <bp@suse.de> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: linux-edac <linux-edac@vger.kernel.org> Link: http://lkml.kernel.org/r/20170327093304.10683-5-bp@alien8.de Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- Jan 24, 2017
-
-
Borislav Petkov authored
Change dependency to mce.c as we're using mce_inject_log() now to stick an MCE into the MCA subsystem. Signed-off-by:
-
- Nov 08, 2016
-
-
Borislav Petkov authored
The MCE injection code does not provide the time stamp information for the injected MCE. Add it. Signed-off-by:
Thomas Gleixner <tglx@linutronix.de>
-
- Sep 26, 2016
-
-
Borislav Petkov authored
Simplify exit_mce_inject() by using debugfs_remove_recursive() and do away with the noodling over the dentry elements. Signed-off-by:
-
Colin Ian King authored
Change predecrement compare to post decrement compare to avoid an unsigned integer wrap-around comparisomn when decrementing in the while loop. For example, if the debugfs_create_file() fails when 'i' is zero, the current situation will predecrement 'i' in the while loop, wrapping 'i' to the maximum signed integer and cause multiple out of bounds reads on dfs_fls[i].d as the loop interates to zero. Also, as Borislav Petkov suggested, return -ENODEV rather than -ENOMEM on the error condition. Signed-off-by:
Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
- Sep 13, 2016
-
-
Borislav Petkov authored
In particular: arch/x86/ras/mce_amd_inj.c: In function ‘prepare_msrs’: arch/x86/ras/mce_amd_inj.c:249:13: warning: declaration of ‘i_mce’ shadows a global declaration [-Wshadow] struct mce i_mce = *(struct mce *)info; ^~~~~ arch/x86/ras/mce_amd_inj.c: In function ‘init_mce_inject’: arch/x86/ras/mce_amd_inj.c:453:16: warning: comparison between signed and unsigned integer expressions [-Wsign-compare] for (i = 0; i < ARRAY_SIZE(dfs_fls); i++) { Signed-off-by: -
Yazen Ghannam authored
Bank 4 is reserved on family 0x17 and shouldn't generate any MCE records. However, broken hardware and software is not something unheard of so warn about bank 4 errors. They shouldn't be coming from bank 4 naturally but users can still use mce_amd_inj to simulate errors from it for testing purposed. Also, avoid special handling in the injector mce_amd_inj like it is being done on the older families. [ bp: Rewrite commit message and merge into one patch. Use boot_cpu_data. ] Signed-off-by:
Yazen Ghannam <Yazen.Ghannam@amd.com> Signed-off-by:
Aravind Gopalakrishnan <aravindksg.lkml@gmail.com> Link: http://lkml.kernel.org/r/1473384591-5323-1-git-send-email-Yazen.Ghannam@amd.com Link: http://lkml.kernel.org/r/1473384591-5323-2-git-send-email-Yazen.Ghannam@amd.com Signed-off-by:
-
Yazen Ghannam authored
Add a debugfs file which holds the error syndrome (written into MCA_SYND) of an injected error. Only write it on SMCA systems. Update README file, while at it. Signed-off-by:
-
- Jul 08, 2016
-
-
Yazen Ghannam authored
We currently use wrmsr_on_cpu() 4 times when prepping for an error injection. This will generate 4 IPIs for each MSR write. We can reduce the number of IPIs to 1 by grouping the MSR writes and executing them serially on the appropriate CPU. Suggested-by:
-
- May 12, 2016
-
-
Yazen Ghannam authored
Use SMCA MSRs when writing to MCA_{STATUS,ADDR,MISC} and MCA_DE{STAT,ADDR} when injecting Deferred Errors on SMCA platforms. Signed-off-by:
-
- Apr 13, 2016
-
-
Borislav Petkov authored
... to be the same like the file name of injection module itself to avoid confusion when grepping. No functionality change. Signed-off-by:
-
- Mar 29, 2016
-
-
Peter Zijlstra authored
It turns out AMD gets x86_max_cores wrong when there are compute units. The issue is that Linux assumes: nr_logical_cpus = nr_cores * nr_siblings But AMD reports its CU unit as 2 cores, but then sets num_smp_siblings to 2 as well. Boris: fixup ras/mce_amd_inj.c too, to compute the Node Base Core properly, according to the new nomenclature. Fixes: 1f12e32f ("x86/topology: Create logical package id") Reported-by:
Xiong Zhou <jencce.kernel@gmail.com> Signed-off-by:
Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by:
-
- Oct 12, 2015
-
-
Aravind Gopalakrishnan authored
Bank 4 MCEs are logged and reported only on the node base core (NBC) in a socket. Refer to the D18F3x44[NbMcaToMstCpuEn] field in Fam10h and later BKDGs. The node base core (NBC) is the lowest numbered core in the node. This patch ensures that we inject the error on the NBC for bank 4 errors. Otherwise, triggering #MC or APIC interrupts on a core which is not the NBC would not have any effect on the system, i.e. we would not see any relevant output on kernel logs for the error we just injected. Signed-off-by:
Aravind Gopalakrishnan <Aravind.Gopalakrishnan@amd.com> [ Cleanup comments. ] [ Add a missing dependency on AMD_NB caught by Randy Dunlap. ] Signed-off-by:
Randy Dunlap <rdunlap@infradead.org> Cc: Borislav Petkov <bp@alien8.de> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tony Luck <tony.luck@intel.com> Link: http://lkml.kernel.org/r/1443190851-2172-4-git-send-email-Aravind.Gopalakrishnan@amd.com Link: http://lkml.kernel.org/r/1444641762-9437-5-git-send-email-bp@alien8.de Signed-off-by:
-
Aravind Gopalakrishnan authored
Add the capability to trigger deferred error interrupts and threshold interrupts in order to test the APIC interrupt handler functionality for these type of errors. Update README section about the same too. Reported by: kbuild test robot <fengguang.wu@intel.com> Signed-off-by:
-
Aravind Gopalakrishnan authored
Invalid inputs such as these are currently reported in dmesg as failing: $> echo sweet > flags [ 122.079139] flags_write: Invalid flags value: et even though the 'flags' attribute has been updated correctly: $> cat flags sw This is because userspace keeps writing the remaining buffer until it encounters an error. However, the input as a whole is wrong and we should not be writing anything to the file. Therefore, correct flags_write() to return -EINVAL immediately on bad input strings. Signed-off-by:
-
- Aug 13, 2015
-
-
Borislav Petkov authored
This is an x86-specific module and would benefit from being closer to the arch code. Move it there. Update copyright while at it. Signed-off-by:
-
