- 25 Mar, 2020 2 commits
-
-
Thomas Richter authored
When perf record -e SF_CYCLES_BASIC_DIAG runs with very high frequency, the samples arrive faster than the perf process can save them to file. Eventually, for longer running processes, this leads to the siutation where the trace buffers allocated by perf slowly fills up. At one point the auxiliary trace buffer is full and the CPU Measurement sampling facility is turned off. Furthermore a warning is printed to the kernel log buffer: cpum_sf: The AUX buffer with 0 pages for the diagnostic-sampling mode is full The number of allocated pages for the auxiliary trace buffer is shown as zero pages. That is wrong. Fix this by saving the number of allocated pages before entering the work loop in the interrupt handler. When the interrupt handler processes the samples, it may detect the buffer full condition and stop sampling, reducing the buffer size to zero. Print the correct value in the error message: cpum_sf: The AUX buffer with 256 pages for the diagnostic-sampling mode is full Signed-off-by:
Thomas Richter <tmricht@linux.ibm.com> Signed-off-by:
Vasily Gorbik <gor@linux.ibm.com>
-
Michael Mueller authored
Show the full diag statistic table and not just parts of it. The issue surfaced in a KVM guest with a number of vcpus defined smaller than NR_DIAG_STAT. Fixes: 1ec2772e ("s390/diag: add a statistic for diagnose calls") Cc: stable@vger.kernel.org Signed-off-by:
Michael Mueller <mimu@linux.ibm.com> Reviewed-by:
Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by:
-
- 23 Mar, 2020 17 commits
-
-
Harald Freudenberger authored
With the removal of the s390 hibernate support the suspend and resume callbacks for the ap devices are not needed any more. This patch removes the callbacks and the ap bus' registration struct for the power management. Signed-off-by:
Harald Freudenberger <freude@linux.ibm.com> Signed-off-by:
-
Niklas Schnelle authored
When UID checking is enabled a UID value of 0 is invalid and can not be set by the user. On z/VM it is however used to indicate an unset UID. Until now, this lead to the behavior that one PCI function could be attached with UID 0 after which z/VM would prohibit further attachment. Now if the user then turns off UID checking in z/VM the user could seemingly attach additional PCI functions that would however not show up in Linux as that would not be informed of the change in UID checking mode. This is unexpected and confusing and lead to bug reports against Linux. Instead now, if we encounter an unset UID value of 0 treat it as indicating that UID checking was turned off, switch to automatic domain allocation, and warn the user of the possible misconfiguration. Signed-off-by:
Niklas Schnelle <schnelle@linux.ibm.com> Reviewed-by:
Peter Oberparleiter <oberpar@linux.ibm.com> Signed-off-by:
-
Niklas Schnelle authored
Until now zpci_alloc_domain() only prevented more than CONFIG_PCI_NR_FUNCTIONS from being added when using automatic domain allocation. When explicit UIDs were defined UIDs above CONFIG_PCI_NR_FUNCTIONS were not counted at all. When more PCI functions are added this could lead to various errors including under sized IRQ vectors and similar issues. Fix this by explicitly tracking the number of allocated domains. Signed-off-by:
Pierre Morel <pmorel@linux.ibm.com> Signed-off-by:
-
Julian Wiedmann authored
When issuing a SADC for a QDIO device, don't hardcode the ISC but use whatever is specified in qdio's handler for Adapter Interrupts. Signed-off-by:
Julian Wiedmann <jwi@linux.ibm.com> Reviewed-by:
Benjamin Block <bblock@linux.ibm.com> Signed-off-by:
-
Heiko Carstens authored
ARCH_SAVE_PAGE_KEYS has been introduced in order to be able to save and restore s390 specific storage keys into a hibernation image. With hibernation support removed from s390 there is no point in keeping the callbacks. Acked-by:
Christian Borntraeger <borntraeger@de.ibm.com> Acked-by:
-
Heiko Carstens authored
Hibernation is known to be broken for many years on s390. Given that there aren't any real use cases, remove the code instead of spending time to fix and maintain it. Without hibernate support it doesn't make too much sense to keep power management support; therefore remove it completely. Acked-by:
-
Alexander Gordeev authored
In the past there were no per-CPU information in /proc/cpuinfo other than CPU frequency. Hence, for machines without CPU MHz feature there were nothing to show. Now CPU topology and IDs still could be shown, so do not skip this information from the output. Signed-off-by:
Alexander Gordeev <agordeev@linux.ibm.com> Reviewed-by:
-
Alexander Gordeev authored
/proc/cpuinfo should not print information about CPU 0 when it is offline. Fixes: 281eaa8c ("s390/cpuinfo: simplify locking and skip offline cpus early") Signed-off-by:
-
Alexander Gordeev authored
Show number of online CPUs within a package (which is the socket in case of s390). For what it worth, present that value as "siblings" field - just like x86 does. Signed-off-by:
-
Alexander Gordeev authored
Show number of cores that run at least one SMT thread Signed-off-by:
-
Gerald Schaefer authored
Re-IPL for both CCW and FCP is currently done by using diag 308 with the "Load Clear" subcode, which means that all memory will be cleared. This can increase re-IPL duration considerably on very large machines. For CCW devices, there is also a "Load Normal" subcode that was only used for dump kernels so far. For FCP devices, a similar "Load Normal" subcode was introduced with z14. The "Load Normal" diag 308 subcode allows to re-IPL without clearing memory. This patch adds a new "clear" sysfs attribute to /sys/firmware/reipl for both the ccw and fcp subdirectories, which can be set to either "0" or "1" to disable or enable re-IPL with memory clearing. The default value is "0", which disables memory clearing. Signed-off-by:
Gerald Schaefer <gerald.schaefer@de.ibm.com> Reviewed-by:
-
Harald Freudenberger authored
snprintf() may not always return the correct size of used bytes but instead the length the resulting string would be if it would fit into the buffer. So scnprintf() is the function to use when the real length of the resulting string is needed. Replace all occurrences of snprintf() with scnprintf() where the return code is further processed. Also find and fix some occurrences where sprintf() was used. Suggested-by:
Takashi Iwai <tiwai@suse.de> Signed-off-by:
-
Takashi Iwai authored
Since snprintf() returns the would-be-output size instead of the actual output size, the succeeding calls may go beyond the given buffer limit. Fix it by replacing with scnprintf(). Message-Id: <20200311090915.21059-1-tiwai@suse.de> Signed-off-by:
-
Alexander Gordeev authored
The CPU topology masks on s390 contain also bits of CPUs which are offline. Currently this is already a problem, since common code scheduler expects e.g. cpu_smt_mask() to reflect reality. This update changes the described behaviour and s390 starts to behave like all other architectures. Signed-off-by:
-
Alexander Gordeev authored
Variable cpus_with_topology is a leftover that became unneeded once the fake NUMA support has been removed. Signed-off-by:
-
Alexander Gordeev authored
Show CPU physical address as reported by STAP instruction Signed-off-by:
-
Alexander Gordeev authored
There is a maximum of two new tables allocated on page table upgrade. Because we know that a loop the current implementation is based on could be unrolled with some improvements: * upgrade from 3 to 5 levels happens in one go - without an unnecessary re-take of page_table_lock in-between; * page tables initialization moved out of the atomic code; Signed-off-by:
-
- 11 Mar, 2020 4 commits
-
-
Christian Borntraeger authored
select does not ensure that dependencies are also selected. Instead of selecting VIRTIO_CONSOLE from S390_GUEST we should rather add this to the defconfigs. So we update those as well. Reported-by:
Michael S. Tsirkin <mst@redhat.com> Signed-off-by:
-
Heiko Carstens authored
Signed-off-by:
-
Heiko Carstens authored
Signed-off-by:
-
afzal mohammed authored
request_irq() is preferred over setup_irq(). Invocations of setup_irq() occur after memory allocators are ready. Per tglx[1], setup_irq() existed in olden days when allocators were not ready by the time early interrupts were initialized. Hence replace setup_irq() by request_irq(). [1] https://lkml.kernel.org/r/alpine.DEB.2.20.1710191609480.1971@nanosSigned-off-by:
afzal mohammed <afzal.mohd.ma@gmail.com> Message-Id: <20200304005049.5291-1-afzal.mohd.ma@gmail.com> [heiko.carstens@de.ibm.com: replace pr_err with panic] Signed-off-by:
-
- 10 Mar, 2020 5 commits
-
-
Alexander Gordeev authored
This update adjusts /proc/cpuinfo format to meet some user level programs expectations. It also makes the layout consistent with x86 where CPU topology is presented as blocks of key-value pairs. Reviewed-by:
-
Sven Schnelle authored
When userspace executes a syscall or gets interrupted, BEAR contains a kernel address when returning to userspace. This make it pretty easy to figure out where the kernel is mapped even with KASLR enabled. To fix this, add lpswe to lowcore and always execute it there, so userspace sees only the lowcore address of lpswe. For this we have to extend both critical_cleanup and the SWITCH_ASYNC macro to also check for lpswe addresses in lowcore. Fixes: b2d24b97 ("s390/kernel: add support for kernel address space layout randomization (KASLR)") Cc: <stable@vger.kernel.org> # v5.2+ Reviewed-by:
Sven Schnelle <svens@linux.ibm.com> Signed-off-by:
-
Heiko Carstens authored
This is the s390 variant of commit 81c22041 ("bpf, x86, arm64: Enable jit by default when not built as always-on"). Acked-by:
Ilya Leoshkevich <iii@linux.ibm.com> Signed-off-by:
-
Thomas Richter authored
Add CPU measurement counter facility event description for IBM z15. Signed-off-by:
Sumanth Korikkar <sumanthk@linux.ibm.com> Signed-off-by:
-
Pierre Morel authored
Embedding the hotplug_slot in zdev structure allows to greatly simplify the hotplug handling by eliminating the handling of the slot_list. Signed-off-by:
-
- 06 Mar, 2020 2 commits
-
-
Julian Wiedmann authored
While we print out various SSQD fields at initialization time, having raw & full access to the current SSQD can help with debugging. Signed-off-by:
-
Alexander Gordeev authored
Make page, frame, virtual and physical address conversion macros more expressive by avoiding redundant definitions and defining new macros using existing ones. Signed-off-by:
-
- 04 Mar, 2020 3 commits
-
-
Peter Oberparleiter authored
Update maintainer entry for s390 PCI subsystem. Sebastian has left IBM, Niklas takes over maintainership. Signed-off-by:
-
Peter Oberparleiter authored
Update maintainer entry for s390 Common I/O Layer. Sebastian has left IBM, Vineeth takes over maintainership. Signed-off-by:
Vineeth Vijayan <vneethv@linux.ibm.com> Signed-off-by:
-
Christian Borntraeger authored
This fixes several sparse warnings for fault.c: arch/s390/mm/fault.c:336:36: warning: restricted vm_fault_t degrades to integer arch/s390/mm/fault.c:573:23: warning: incorrect type in assignment (different base types) arch/s390/mm/fault.c:573:23: expected restricted vm_fault_t [usertype] fault arch/s390/mm/fault.c:573:23: got int Signed-off-by:
-
- 27 Feb, 2020 5 commits
-
-
Heiko Carstens authored
It turned out that fake numa support is rather useless on s390, since there are no scenarios where there is any performance or other benefit when used. However it does provide maintenance cost and breaks from time to time. Therefore remove it. CONFIG_NUMA is still supported with a very small backend and only one node. This way userspace applications which require NUMA interfaces continue to work. Note that NODES_SHIFT is set to 1 (= 2 nodes) instead of 0 (= 1 node), since there is quite a bit of kernel code which assumes that more than one node is possible if CONFIG_NUMA is enabled. Signed-off-by:
-
Torsten Duwe authored
aes_s390.c has several functions which allocate space for key material on the stack and leave the used keys there. It is considered good practice to clean these locations before the function returns. Link: https://lkml.kernel.org/r/20200221165511.GB6928@lst.deSigned-off-by:
Torsten Duwe <duwe@suse.de> Signed-off-by:
-
Gustavo A. R. Silva authored
The current codebase makes use of the zero-length array language extension to the C90 standard, but the preferred mechanism to declare variable-length types such as these ones is a flexible array member[1][2], introduced in C99: struct foo { int stuff; struct boo array[]; }; By making use of the mechanism above, we will get a compiler warning in case the flexible array does not occur last in the structure, which will help us prevent some kind of undefined behavior bugs from being inadvertently introduced[3] to the codebase from now on. Also, notice that, dynamic memory allocations won't be affected by this change: "Flexible array members have incomplete type, and so the sizeof operator may not be applied. As a quirk of the original implementation of zero-length arrays, sizeof evaluates to zero."[1] This issue was found with the help of Coccinelle. [1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html [2] https://github.com/KSPP/linux/issues/21 [3] commit 76497732 ("cxgb3/l2t: Fix undefined behaviour") Link: https://lkml.kernel.org/r/20200221150612.GA9717@embeddedorSigned-off-by:Gustavo A. R. Silva <gustavo@embeddedor.com> Signed-off-by:
-
Julian Wiedmann authored
There's no need for error handling, the debugfs core is smart enough to deal with IS_ERR() internally. This will also keep us from creating the debugfs files if the device directory doesn't exist. Currently (because irq_ptr->debugfs_dev gets set to NULL on error) the files would be placed into the debugfs root - without any association to their parent device. On teardown, use the debugfs_remove_recursive() helper to avoid keeping track of each created file/directory. Signed-off-by:
-
Julian Wiedmann authored
Don't rely on the numeric value of enum constants. Signed-off-by:
-
- 19 Feb, 2020 2 commits
-
-
Stefan Raspl authored
In case $INSTALLKERNEL is not available, we should install the kernel image with its version number, and save the previous one accordingly. Also, we're adding a hint so users know that they still need to perform one more configuration step (usually adjusting zipl config). Signed-off-by:
Stefan Raspl <raspl@linux.ibm.com> Signed-off-by:
-
Julian Wiedmann authored
Remove all usage of cdev->private->qdio_data that's buried deep in internal code. This should only be used by the exported driver API, which can then pass around a proper qdio_irq pointer. Also trivially merge some initializations with their definitions. Signed-off-by:
-
