- 20 Nov, 2019 6 commits
-
-
Heiko Carstens authored
If an SMT capable system is not IPL'ed from the first CPU the setup of the physical to logical CPU mapping is broken: the IPL core gets CPU number 0, but then the next core gets CPU number 1. Correct would be that all SMT threads of CPU 0 get the subsequent logical CPU numbers. This is important since a lot of code (like e.g. the CPU topology code) assumes that CPU maps are setup like this. If the mapping is broken the system will not IPL due to broken topology masks: [ 1.716341] BUG: arch topology broken [ 1.716342] the SMT domain not a subset of the MC domain [ 1.716343] BUG: arch topology broken [ 1.716344] the MC domain not a subset of the BOOK domain This scenario can usually not happen since LPARs are always IPL'ed from CPU 0 and also re-IPL is intiated from CPU 0. However older kernels did initiate re-IPL on an arbitrary CPU. If therefore a re-IPL from an old kernel into a new kernel is initiated this may lead to crash. Fix this by setting up the physical to logical CPU mapping correctly. Signed-off-by:
Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by:
Vasily Gorbik <gor@linux.ibm.com>
-
Vasily Gorbik authored
Reviewed-by:
-
Vasily Gorbik authored
vdso_per_cpu_data lowcore value is only needed for fully functional exception handlers, which are activated in setup_lowcore_dat_off. The same function does init vdso_per_cpu_data via vdso_alloc_boot_cpu. Reviewed-by:
-
Vasily Gorbik authored
Reviewed-by:
-
Vasily Gorbik authored
Currently if the kernel is built with CONFIG_TRACE_IRQFLAGS and KASAN and used as crash kernel it crashes itself due to trace_hardirqs_off/trace_hardirqs_on being called with DAT off. This happens because trace_hardirqs_off/trace_hardirqs_on are instrumented and kasan code tries to perform access to shadow memory to validate memory accesses. Kasan shadow memory is populated with vmemmap, so all accesses require DAT on. memcpy_real could be called with DAT on or off (with kasan enabled DAT is set even before early code is executed). Make sure that trace_hardirqs_off/trace_hardirqs_on are called with DAT on and only actual __memcpy_real is called with DAT off. Also annotate __memcpy_real and _memcpy_real with __no_sanitize_address to avoid further problems due to switching DAT off. Reviewed-by:
Philipp Rudo <prudo@linux.ibm.com> Signed-off-by:
-
YueHaibing authored
s390_crypto_shash_parmsize() return type is int, it should not be stored in a unsigned variable, which compared with zero. Reported-by:
Hulk Robot <hulkci@huawei.com> Fixes: 3c2eb6b7 ("s390/crypto: Support for SHA3 via CPACF (MSA6)") Signed-off-by:
YueHaibing <yuehaibing@huawei.com> Signed-off-by:
Joerg Schmidbauer <jschmidb@linux.vnet.ibm.com> Signed-off-by:
-
- 12 Nov, 2019 7 commits
-
-
Markus Elfring authored
Generated by: scripts/coccinelle/api/memdup_user.cocci Link: http://lkml.kernel.org/r/aca044e8-e4b2-eda8-d724-b08772a44ed9@web.de [borntraeger@de.ibm.com: use ==0 instead of <=0 for a size_t variable] [heiko.carstens@de.ibm.com: split bugfix into separate patch; shorten changelog] Signed-off-by:
Markus Elfring <Markus.Elfring@web.de> Signed-off-by:
Christian Borntraeger <borntraeger@de.ibm.com> Signed-off-by:
-
Heiko Carstens authored
Fixes: f2bbc96e ("s390/pkey: add CCA AES cipher key support") Reported-by:
-
Vasily Gorbik authored
Merge tag 'vfio-ccw-20191111' of https://git.kernel.org/pub/scm/linux/kernel/git/kvms390/vfio-ccw into features enhance tracing in vfio-ccw * tag 'vfio-ccw-20191111' of https://git.kernel.org/pub/scm/linux/kernel/git/kvms390/vfio-ccw: vfio-ccw: Rework the io_fctl trace vfio-ccw: Add a trace for asynchronous requests vfio-ccw: Trace the FSM jumptable vfio-ccw: Refactor how the traces are built Signed-off-by:
-
Ilya Leoshkevich authored
Due to kptr_restrict, JITted BPF code is now displayed like this: 000000000b6ed1b2: ebdff0800024 stmg %r13,%r15,128(%r15) 000000004cde2ba0: 41d0f040 la %r13,64(%r15) 00000000fbad41b0: a7fbffa0 aghi %r15,-96 Leaking kernel addresses to dmesg is not a concern in this case, because this happens only when JIT debugging is explicitly activated, which only root can do. Use %px in this particular instance, and also to print an instruction address in show_code and PCREL (e.g. brasl) arguments in print_insn. While at present functionally equivalent to %016lx, %px is recommended by Documentation/core-api/printk-formats.rst for such cases. Signed-off-by:
Ilya Leoshkevich <iii@linux.ibm.com> Reviewed-by:
-
Thomas Richter authored
When starting the CPU Measurement sampling facility using qsi() function, this function may return an error value. This error value is referenced in the else part of the if statement to dump its value in a debug statement. Right now this value is always zero because it has not been assigned a value. Signed-off-by:
Thomas Richter <tmricht@linux.ibm.com> Signed-off-by:
-
Thomas Richter authored
Replace hard coded function names in debug statements by the "%s ...", __func__ construct suggested by checkpatch.pl script. Signed-off-by:
-
Thomas Richter authored
Use consistant debug print format of the form variable blank value. Also add leading 0x for all hex values. Signed-off-by:
-
- 05 Nov, 2019 1 commit
-
-
Miroslav Benes authored
The current code around calling ftrace_graph_ret_addr() is ifdeffed and also tests if ftrace redirection is present on stack. ftrace_graph_ret_addr() however performs the test internally and there is a version for !CONFIG_FUNCTION_GRAPH_TRACER as well. The unnecessary code can thus be dropped. Link: http://lkml.kernel.org/r/20191029143904.24051-2-mbenes@suse.czSigned-off-by:
Miroslav Benes <mbenes@suse.cz> Signed-off-by:
-
- 31 Oct, 2019 21 commits
-
-
Ilya Leoshkevich authored
perf_callchain_kernel stops neither when it encounters a garbage address, nor when it runs out of space. Fix both issues using x86 version as an inspiration. Signed-off-by:
-
Heiko Carstens authored
This function must be inlined since any caller expects the current stack pointer; which wouldn't be true if the function isn't inlined. Signed-off-by:
-
Gerald Schaefer authored
Unlike pxd_free_tlb(), the pxd_free() functions do not check for folded page tables. This is not an issue so far, as those functions will actually never be called, since no code will reach them when page tables are folded. In order to avoid future issues, and to make the s390 code more similar to other architectures, add mm_pxd_folded() checks, similar to how it is done in pxd_free_tlb(). This was found by testing a patch from from Anshuman Khandual, which is currently discussed on LKML ("mm/debug: Add tests validating architecture page table helpers"). Signed-off-by:Gerald Schaefer <gerald.schaefer@de.ibm.com> Signed-off-by:
-
Gerald Schaefer authored
On older HW or under a hypervisor, w/o the instruction-execution- protection (IEP) facility, and also w/o EDAT-1, a translation-specification exception may be recognized when bit 55 of a pte is one (_PAGE_NOEXEC). The current code tries to prevent setting _PAGE_NOEXEC in such cases, by removing it within set_pte_at(). However, ptep_set_access_flags() will modify a pte directly, w/o using set_pte_at(). There is at least one scenario where this can result in an active pte with _PAGE_NOEXEC set, which would then lead to a panic due to a translation-specification exception (write to swapped out page): do_swap_page pte = mk_pte (with _PAGE_NOEXEC bit) set_pte_at (will remove _PAGE_NOEXEC bit in page table, but keep it in local variable pte) vmf->orig_pte = pte (pte still contains _PAGE_NOEXEC bit) do_wp_page wp_page_reuse entry = vmf->orig_pte (still with _PAGE_NOEXEC bit) ptep_set_access_flags (writes entry with _PAGE_NOEXEC bit) Fix this by clearing _PAGE_NOEXEC already in mk_pte_phys(), where the pgprot value is applied, so that no pte with _PAGE_NOEXEC will ever be visible, if it is not supported. The check in set_pte_at() can then also be removed. Cc: <stable@vger.kernel.org> # 4.11+ Fixes: 57d7f939 ("s390: add no-execute support") Signed-off-by: -
Gerald Schaefer authored
For pmds and puds, there are a couple of page table helper functions that only make sense for large entries, like pxd_(mk)dirty/young/write etc. We currently explicitly check if the entries are large, but in practice those functions must never be used for normal entries, which point to lower level page tables, so the code can be simplified. This also fixes a theoretical bug, where common code could use one of the functions before actually marking a pmd large, like this: pmd = pmd_mkhuge(pmd_mkdirty(pmd)) With the current implementation, the resulting large pmd would not be dirty as requested. This could in theory result in the loss of dirty information, e.g. after collapsing into a transparent hugepage. Common code currently always marks an entry large before using one of the functions, but there is no hard requirement for this. The only requirement would be that it never uses the functions for normal entries pointing to lower level page tables, but they might be called before marking an entry large during its creation. In order to avoid issues with future common code, and to simplify the page table helpers, remove the checks for large entries and rely on common code never using them for normal entries. This was found by testing a patch from from Anshuman Khandual, which is currently discussed on LKML ("mm/debug: Add tests validating architecture page table helpers"). Signed-off-by: -
Gerald Schaefer authored
The semantics of pmd/pud_bad() expect that large entries are reported as bad, but we also check large entries for sanity. There is currently no issue with this wrong behaviour, but let's conform to the semantics by reporting large pmd/pud entries as bad, in order to prevent future issues. This was found by testing a patch from from Anshuman Khandual, which is currently discussed on LKML ("mm/debug: Add tests validating architecture page table helpers"). Signed-off-by: -
Heiko Carstens authored
The current implementation of get_clock_monotonic() leaves it up to the caller to call the function with preemption disabled. The only core kernel caller (sched_clock) however does not disable preemption. In order to make sure that all callers of this function see monotonic values handle disabling preemption within the function itself. Signed-off-by:
-
Vasily Gorbik authored
Currently get_wchan uses custom stack unwinding implementation which relies on back_chain presence. Replace it with more abstract stack unwinding api usage. Suggested-by:
-
Ilya Leoshkevich authored
unwind_for_each_frame(NULL, NULL, 0) does not return any valid frames. The reason is that get_stack_pointer, unlike get_stack_info and show_stack, does not handle NULL argument. Fix by making get_stack_pointer treat NULL as current, like get_stack_info and show_stack do. Reviewed-by:
-
Vasily Gorbik authored
"noexec" option is already parsed during startup and its value is exposed via noexec_disabled variable. Simply reuse that value during machine facilities detection. Suggested-by:
-
Heiko Carstens authored
Remove unused monotonic_clock() function. Signed-off-by:
-
Julian Wiedmann authored
Put the Sniffer bit next to all the other CHSC AC2 bits. Signed-off-by:
Julian Wiedmann <jwi@linux.ibm.com> Reviewed-by:
Steffen Maier <maier@linux.ibm.com> Signed-off-by:
-
Nick Desaulniers authored
GCC unescapes escaped string section names while Clang does not. Because __section uses the `#` stringification operator for the section name, it doesn't need to be escaped. This antipattern was found with: $ grep -e __section\(\" -e __section__\(\" -r Reported-by:
Sedat Dilek <sedat.dilek@gmail.com> Suggested-by:
Josh Poimboeuf <jpoimboe@redhat.com> Signed-off-by:
Nick Desaulniers <ndesaulniers@google.com> Message-Id: <20190812215052.71840-1-ndesaulniers@google.com> Signed-off-by:
-
Heiko Carstens authored
This is the s390 version of commit 40576e5e ("x86: alternative.h: use asm_inline for all alternative variants"). See commit eb111869 ("compiler-types.h: add asm_inline definition") for more details. With this change the compiler will not generate many out-of-line versions for the three instruction sized arch_spin_unlock() function anymore. Due to this gcc seems to change a lot of other inline decisions which results in a net 6k text size growth according to bloat-o-meter (gcc 9.2 with defconfig). But that's still better than having many out-of-line versions of arch_spin_unlock(). Signed-off-by:
-
Heiko Carstens authored
This is the s390 version of commit 32ee8230 ("x86: bug.h: use asm_inline in _BUG_FLAGS definitions"). See commit eb111869 ("compiler-types.h: add asm_inline definition") for more details. Just like on x86 the .text section size decreases a bit while the .data section size increases about the same amount (gcc 9.2 with defconfig). Signed-off-by:
-
Julian Wiedmann authored
Output interrupts are not subject to SLSB-based avoidance, so remove the gratuitous SLSB updates for Output SBALs in ERROR state. Signed-off-by:
Benjamin Block <bblock@linux.ibm.com> Signed-off-by:
-
Julian Wiedmann authored
On an interrupt, tiqdio_thinint_handler() walks a list of all objects that might require attention, and checks their DSCI. This list is awkwardly built from Input Queues, even though the IRQs are per-device and the queue is then only used to dereference its qdio_irq parent. To simplify the logic, change the code so that tiq_list contains qdio_irq entries. Signed-off-by:
-
Julian Wiedmann authored
qperf_inc() takes a queue as input, but actually updates the statistics in its qdio_irq parent. In some contexts we already have access to the qdio_irq struct, and can avoid the additional dereference. Signed-off-by:
-
Julian Wiedmann authored
Shift the definition of tiqdio_airq around, so that it doesn't require a forward declaration for tiqdio_thinint_handler(). Signed-off-by:
-
Julian Wiedmann authored
Partial EQBS completion is no significant event, and the WARN ends up spamming the debug logs for no good reason. Signed-off-by:
-
Julian Wiedmann authored
qdio.h recently gained a new helper macro that handles wrap-around on a QDIO queue, use it. Signed-off-by:
-
- 17 Oct, 2019 4 commits
-
-
Eric Farman authored
Using __field_struct for the schib is convenient, but it doesn't appear to let us filter based on any of the schib elements. Specifying the full schid or any element within it results in various errors by the parser. So, expand that out to its component elements, so we can limit the trace to a single device. While we are at it, rename this trace to the function name, so we remember what is being traced instead of an abstract reference to the function control bit of the SCSW. Signed-off-by:
Eric Farman <farman@linux.ibm.com> Reviewed-by:
Cornelia Huck <cohuck@redhat.com> Message-Id: <20191016142040.14132-5-farman@linux.ibm.com> Acked-by:
Halil Pasic <pasic@linux.ibm.com> Signed-off-by:
-
Eric Farman authored
Since the asynchronous requests are typically associated with error recovery, let's add a simple trace when one of those is issued to a device. Signed-off-by:
-
Eric Farman authored
It would be nice if we could track the sequence of events within vfio-ccw, based on the state of the device/FSM and our calling sequence within it. So let's add a simple trace here so we can watch the states change as things go, and allow it to be folded into the rest of the other cio traces. Signed-off-by:
-
Eric Farman authored
Commit 3cd90214 ("vfio: ccw: add tracepoints for interesting error paths") added a quick trace point to determine where a channel program failed while being processed. It's a great addition, but adding more traces to vfio-ccw is more cumbersome than it needs to be. Let's refactor how this is done, so that additional traces are easier to add and can exist outside of the FSM if we ever desire. Signed-off-by:
-
- 10 Oct, 2019 1 commit
-
-
Heiko Carstens authored
The names for the z13s and z14 ZR1 machines are missing for the TUNE_Z13 and TUNE_Z14 descriptions. Just add them. Signed-off-by:
-
