- 25 Apr, 2019 1 commit
-
-
Martin Schwidefsky authored
Merge tag 'vfio-ccw-20190425' of https://git.kernel.org/pub/scm/linux/kernel/git/kvms390/vfio-ccw into features Pull vfio-ccw from Cornelia Huck with the following changes: - support for sending halt/clear requests to the device - various bug fixes
-
- 24 Apr, 2019 11 commits
-
-
Farhan Ali authored
The quiesce function calls cio_cancel_halt_clear() and if we get an -EBUSY we go into a loop where we: - wait for any interrupts - flush all I/O in the workqueue - retry cio_cancel_halt_clear During the period where we are waiting for interrupts or flushing all I/O, the channel subsystem could have completed a halt/clear action and turned off the corresponding activity control bits in the subchannel status word. This means the next time we call cio_cancel_halt_clear(), we will again start by calling cancel subchannel and so we can be stuck between calling cancel and halt forever. Rather than calling cio_cancel_halt_clear() immediately after waiting, let's try to disable the subchannel. If we succeed in disabling the subchannel then we know nothing else can happen with the device. Suggested-by:
Eric Farman <farman@linux.ibm.com> Signed-off-by:
Farhan Ali <alifm@linux.ibm.com> Message-Id: <4d5a4b98ab1b41ac6131b5c36de18b76c5d66898.1555449329.git.alifm@linux.ibm.com> Reviewed-by:
Halil Pasic <pasic@linux.ibm.com> Signed-off-by:
Cornelia Huck <cohuck@redhat.com>
-
Farhan Ali authored
When releasing the vfio-ccw mdev, we currently do not release any existing channel program and its pinned pages. This can lead to the following warning: [1038876.561565] WARNING: CPU: 2 PID: 144727 at drivers/vfio/vfio_iommu_type1.c:1494 vfio_sanity_check_pfn_list+0x40/0x70 [vfio_iommu_type1] .... 1038876.561921] Call Trace: [1038876.561935] ([<00000009897fb870>] 0x9897fb870) [1038876.561949] [<000003ff8013bf62>] vfio_iommu_type1_detach_group+0xda/0x2f0 [vfio_iommu_type1] [1038876.561965] [<000003ff8007b634>] __vfio_group_unset_container+0x64/0x190 [vfio] [1038876.561978] [<000003ff8007b87e>] vfio_group_put_external_user+0x26/0x38 [vfio] [1038876.562024] [<000003ff806fc608>] kvm_vfio_group_put_external_user+0x40/0x60 [kvm] [1038876.562045] [<000003ff806fcb9e>] kvm_vfio_destroy+0x5e/0xd0 [kvm] [1038876.562065] [<000003ff806f63fc>] kvm_put_kvm+0x2a4/0x3d0 [kvm] [1038876.562083] [<000003ff806f655e>] kvm_vm_release+0x36/0x48 [kvm] [1038876.562098] [<00000000003c2dc4>] __fput+0x144/0x228 [1038876.562113] [<000000000016ee82>] task_work_run+0x8a/0xd8 [1038876.562125] [<000000000014c7a8>] do_exit+0x5d8/0xd90 [1038876.562140] [<000000000014d084>] do_group_exit+0xc4/0xc8 [1038876.562155] [<000000000015c046>] get_signal+0x9ae/0xa68 [1038876.562169] [<0000000000108d66>] do_signal+0x66/0x768 [1038876.562185] [<0000000000b9e37e>] system_call+0x1ea/0x2d8 [1038876.562195] 2 locks held by qemu-system-s39/144727: [1038876.562205] #0: 00000000537abaf9 (&container->group_lock){++++}, at: __vfio_group_unset_container+0x3c/0x190 [vfio] [1038876.562230] #1: 00000000670008b5 (&iommu->lock){+.+.}, at: vfio_iommu_type1_detach_group+0x36/0x2f0 [vfio_iommu_type1] [1038876.562250] Last Breaking-Event-Address: [1038876.562262] [<000003ff8013aa24>] vfio_sanity_check_pfn_list+0x3c/0x70 [vfio_iommu_type1] [1038876.562272] irq event stamp: 4236481 [1038876.562287] hardirqs last enabled at (4236489): [<00000000001cee7a>] console_unlock+0x6d2/0x740 [1038876.562299] hardirqs last disabled at (4236496): [<00000000001ce87e>] console_unlock+0xd6/0x740 [1038876.562311] softirqs last enabled at (4234162): [<0000000000b9fa1e>] __do_softirq+0x556/0x598 [1038876.562325] softirqs last disabled at (4234153): [<000000000014e4cc>] irq_exit+0xac/0x108 [1038876.562337] ---[ end trace 6c96d467b1c3ca06 ]--- Similarly we do not free the channel program when we are removing the vfio-ccw device. Let's fix this by resetting the device and freeing the channel program and pinned pages in the release path. For the remove path we can just quiesce the device, since in the remove path the mediated device is going away for good and so we don't need to do a full reset. Signed-off-by: -
Farhan Ali authored
Currently we call flush_workqueue while holding the subchannel spinlock. But flush_workqueue function can go to sleep, so do not call the function while holding the spinlock. Fixes the following bug: [ 285.203430] BUG: scheduling while atomic: bash/14193/0x00000002 [ 285.203434] INFO: lockdep is turned off. .... [ 285.203485] Preemption disabled at: [ 285.203488] [<000003ff80243e5c>] vfio_ccw_sch_quiesce+0xbc/0x120 [vfio_ccw] [ 285.203496] CPU: 7 PID: 14193 Comm: bash Tainted: G W .... [ 285.203504] Call Trace: [ 285.203510] ([<0000000000113772>] show_stack+0x82/0xd0) [ 285.203514] [<0000000000b7a102>] dump_stack+0x92/0xd0 [ 285.203518] [<000000000017b8be>] __schedule_bug+0xde/0xf8 [ 285.203524] [<0000000000b95b5a>] __schedule+0x7a/0xc38 [ 285.203528] [<0000000000b9678a>] schedule+0x72/0xb0 [ 285.203533] [<0000000000b9bfbc>] schedule_timeout+0x34/0x528 [ 285.203538] [<0000000000b97608>] wait_for_common+0x118/0x1b0 [ 285.203544] [<0000000000166d6a>] flush_workqueue+0x182/0x548 [ 285.203550] [<000003ff80243e6e>] vfio_ccw_sch_quiesce+0xce/0x120 [vfio_ccw] [ 285.203556] [<000003ff80245278>] vfio_ccw_mdev_reset+0x38/0x70 [vfio_ccw] [ 285.203562] [<000003ff802458b0>] vfio_ccw_mdev_remove+0x40/0x78 [vfio_ccw] [ 285.203567] [<000003ff801a499c>] mdev_device_remove_ops+0x3c/0x80 [mdev] [ 285.203573] [<000003ff801a4d5c>] mdev_device_remove+0xc4/0x130 [mdev] [ 285.203578] [<000003ff801a5074>] remove_store+0x6c/0xa8 [mdev] [ 285.203582] [<000000000046f494>] kernfs_fop_write+0x14c/0x1f8 [ 285.203588] [<00000000003c1530>] __vfs_write+0x38/0x1a8 [ 285.203593] [<00000000003c187c>] vfs_write+0xb4/0x198 [ 285.203597] [<00000000003c1af2>] ksys_write+0x5a/0xb0 [ 285.203601] [<0000000000b9e270>] system_call+0xdc/0x2d8 Signed-off-by:
Pierre Morel <pmorel@linux.ibm.com> Message-Id: <626bab8bb2958ae132452e1ddaf1b20882ad5a9d.1554756534.git.alifm@linux.ibm.com> Signed-off-by:
-
Cornelia Huck authored
Add a region to the vfio-ccw device that can be used to submit asynchronous I/O instructions. ssch continues to be handled by the existing I/O region; the new region handles hsch and csch. Interrupt status continues to be reported through the same channels as for ssch. Acked-by:
-
Cornelia Huck authored
The vfio-ccw code will need this, and it matches treatment of ssch and csch. Reviewed-by:
-
Cornelia Huck authored
Allow to extend the regions used by vfio-ccw. The first user will be handling of halt and clear subchannel. Reviewed-by:
-
Cornelia Huck authored
Introduce a mutex to disallow concurrent reads or writes to the I/O region. This makes sure that the data the kernel or user space see is always consistent. The same mutex will be used to protect the async region as well. Reviewed-by:
-
Cornelia Huck authored
The flow for processing ssch requests can be improved by splitting the BUSY state: - CP_PROCESSING: We reject any user space requests while we are in the process of translating a channel program and submitting it to the hardware. Use -EAGAIN to signal user space that it should retry the request. - CP_PENDING: We have successfully submitted a request with ssch and are now expecting an interrupt. As we can't handle more than one channel program being processed, reject any further requests with -EBUSY. A final interrupt will move us out of this state. By making this a separate state, we make it possible to issue a halt or a clear while we're still waiting for the final interrupt for the ssch (in a follow-on patch). It also makes a lot of sense not to preemptively filter out writes to the io_region if we're in an incorrect state: the state machine will handle this correctly. Reviewed-by:
-
Cornelia Huck authored
When we get a solicited interrupt, the start function may have been cleared by a csch, but we still have a channel program structure allocated. Make it safe to call the cp accessors in any case, so we can call them unconditionally. While at it, also make sure that functions called from other parts of the code return gracefully if the channel program structure has not been initialized (even though that is a bug in the caller). Reviewed-by:
-
Martin Schwidefsky authored
With git commit d1874a0c "s390/mm: make the pxd_offset functions more robust" and a 2-level page table it can now happen that pgd_bad() gets asked to verify a large segment table entry. If the entry is marked as dirty pgd_bad() will incorrectly return true. Change the pgd_bad(), p4d_bad(), pud_bad() and pmd_bad() functions to first verify the table type, return false if the table level is lower than what the function is suppossed to check, return true if the table level is too high, and otherwise check the relevant region and segment table bits. pmd_bad() has to check against ~SEGMENT_ENTRY_BITS for normal page table pointers or ~SEGMENT_ENTRY_BITS_LARGE for large segment table entries. Same for pud_bad() which has to check against ~_REGION_ENTRY_BITS or ~_REGION_ENTRY_BITS_LARGE. Fixes: d1874a0c ("s390/mm: make the pxd_offset functions more robust") Signed-off-by:
Martin Schwidefsky <schwidefsky@de.ibm.com>
-
Vasily Gorbik authored
arch/s390/lib/uaccess.c is built without kasan instrumentation. Kasan checks are performed explicitly in copy_from_user/copy_to_user functions. But since those functions could be inlined, calls from files like uaccess.c with instrumentation disabled won't generate kasan reports. This is currently the case with strncpy_from_user function which was revealed by newly added kasan test. Avoid inlining of copy_from_user/copy_to_user when the kernel is built with kasan support to make sure kasan checks are fully functional. Signed-off-by:
Vasily Gorbik <gor@linux.ibm.com> Signed-off-by:
-
- 23 Apr, 2019 3 commits
-
-
Thomas-Mich Richter authored
Add support for the CPU-Measurement Facility counter second version number 6. This number is used to detect some more counters in the crypto counter set and the extended counter set. Signed-off-by:
Thomas Richter <tmricht@linux.ibm.com> Reviewed-by:
Hendrik Brueckner <brueckner@linux.ibm.com> Signed-off-by:
-
Martin Schwidefsky authored
Define the gup_fast_permitted to check against the asce_limit of the mm attached to the current task, then replace the s390 specific gup code with the generic implementation in mm/gup.c. Signed-off-by: -
Martin Schwidefsky authored
Change the way how pgd_offset, p4d_offset, pud_offset and pmd_offset walk the page tables. pgd_offset now always calculates the index for the top-level page table and adds it to the pgd, this is either a segment table offset for a 2-level setup, a region-3 offset for 3-levels, region-2 offset for 4-levels, or a region-1 offset for a 5-level setup. The other three functions p4d_offset, pud_offset and pmd_offset will only add the respective offset if they dereference the passed pointer. With the new way of walking the page tables a sequence like this from mm/gup.c now works: pgdp = pgd_offset(current->mm, addr); pgd = READ_ONCE(*pgdp); p4dp = p4d_offset(&pgd, addr); p4d = READ_ONCE(*p4dp); pudp = pud_offset(&p4d, addr); pud = READ_ONCE(*pudp); pmdp = pmd_offset(&pud, addr); pmd = READ_ONCE(*pmdp); Signed-off-by:
-
- 18 Apr, 2019 2 commits
-
-
Julian Wiedmann authored
qdio.ko offers a small number of high-level functions to drive the scanning of a QDIO queue for ready-to-process SBALs: qdio_get_next_buffers(), __[ti]qdio_inbound_processing() and __qdio_outbound_processing(). Let each of those functions maintain the 'start' index for their current scan, and pass it to lower-level helpers as needed. This improves the code's overall layering, and allows us to eliminate the additional first_to_kick cursor with a follow-on patch. Signed-off-by:
Julian Wiedmann <jwi@linux.ibm.com> Reviewed-by:
Jens Remus <jremus@linux.ibm.com> Signed-off-by:
-
Julian Wiedmann authored
Refactor all the low-level helpers to take the first_to_check cursor as parameter, rather than accessing it directly. Signed-off-by:
-
- 11 Apr, 2019 9 commits
-
-
Arnd Bergmann authored
The 'func_code' variable gets printed in debug statements without a prior initialization in multiple functions, as reported when building with clang: drivers/s390/crypto/zcrypt_api.c:659:6: warning: variable 'func_code' is used uninitialized whenever 'if' condition is true [-Wsometimes-uninitialized] if (mex->outputdatalength < mex->inputdatalength) { ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ drivers/s390/crypto/zcrypt_api.c:725:29: note: uninitialized use occurs here trace_s390_zcrypt_rep(mex, func_code, rc, ^~~~~~~~~ drivers/s390/crypto/zcrypt_api.c:659:2: note: remove the 'if' if its condition is always false if (mex->outputdatalength < mex->inputdatalength) { ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ drivers/s390/crypto/zcrypt_api.c:654:24: note: initialize the variable 'func_code' to silence this warning unsigned int func_code; ^ Add initializations to all affected code paths to shut up the warning and make the warning output consistent. Signed-off-by:Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Arnd Bergmann authored
llvm on s390 has problems with __builtin_return_address(n), with n>0, this results in a somewhat cryptic error message: fatal error: error in backend: Unsupported stack frame traversal count To work around it, use the direct return address directly. This is probably not ideal here, but gets things to compile and should only lead to inferior reporting, not to misbehavior of the generated code. Link: https://bugs.llvm.org/show_bug.cgi?id=41424Signed-off-by:
-
Joe Perches authored
IS_ENABLED should be reserved for CONFIG_<FOO> uses so convert the uses of IS_ENABLED with a #define to __is_defined. Signed-off-by:
Joe Perches <joe@perches.com> Signed-off-by:
-
Arnd Bergmann authored
llvm skips an empty .bss section entirely, which makes the check fail with an unexpected error: /tmp/binutils-multi-test/bin/s390x-linux-gnu-objdump: section '.bss' mentioned in a -j option, but not found in any input file error: arch/s390/boot/compressed/decompressor.o .bss section is not empty ../arch/s390/scripts/Makefile.chkbss:20: recipe for target 'arch/s390/boot/compressed/decompressor.o.chkbss' failed Change the check so we first see if a .bss section exists before trying to read its size. Signed-off-by:
-
Arnd Bergmann authored
clang fails to use the %O and %R inline assembly modifiers the same way as gcc, leading to build failures with every use of __load_psw_mask(): /tmp/nmi-4a9f80.s: Assembler messages: /tmp/nmi-4a9f80.s:571: Error: junk at end of line: `+8(160(%r11))' /tmp/nmi-4a9f80.s:626: Error: junk at end of line: `+8(160(%r11))' Replace these with a more conventional way of passing the addresses that should work with both clang and gcc. Signed-off-by:
-
Arnd Bergmann authored
Building system calls with clang results in a warning about an alias from a global function to a static one: ../fs/namei.c:3847:1: warning: unused function '__se_sys_mkdirat' [-Wunused-function] SYSCALL_DEFINE3(mkdirat, int, dfd, const char __user *, pathname, umode_t, mode) ^ ../include/linux/syscalls.h:219:36: note: expanded from macro 'SYSCALL_DEFINE3' #define SYSCALL_DEFINE3(name, ...) SYSCALL_DEFINEx(3, _##name, __VA_ARGS__) ^ ../include/linux/syscalls.h:228:2: note: expanded from macro 'SYSCALL_DEFINEx' __SYSCALL_DEFINEx(x, sname, __VA_ARGS__) ^ ../arch/s390/include/asm/syscall_wrapper.h:126:18: note: expanded from macro '__SYSCALL_DEFINEx' asmlinkage long __se_sys##name(__MAP(x,__SC_LONG,__VA_ARGS__)) \ ^ <scratch space>:31:1: note: expanded from here __se_sys_mkdirat ^ The only reference to the static __se_sys_mkdirat() here is the alias, but this only gets evaluated later. Making this function global as well avoids the warning. Signed-off-by: -
Arnd Bergmann authored
clang does not support 31 bit object files on s390, so skip the 32-bit vdso here, and only build it when using gcc to compile the kernel. Signed-off-by:
-
Arnd Bergmann authored
This was added as a workaround for really old compilers, and it prevents building with clang now. I can see no reason for keeping it, as it has already been removed for most architectures in the pre-git era, so let's remove it everywhere, rather than only for clang. Signed-off-by:
-
Martin Schwidefsky authored
The CALL_ON_STACK helper currently does not work with clang and for calls without arguments. It does not initialize r2 although the constraint is "+&d". Rework the CALL_FMT_x and the CALL_ON_STACK macros to work with clang and produce optimal code in all cases. Reported-by:
-
- 10 Apr, 2019 13 commits
-
-
Martin Schwidefsky authored
Use trap4 as the guard instruction for the restartable sequence abort handler. Signed-off-by: -
Arnd Bergmann authored
clang points out that the declaration of cio_irb does not match the definition exactly, it is missing the alignment attribute: ../drivers/s390/cio/cio.c:50:1: warning: section does not match previous declaration [-Wsection] DEFINE_PER_CPU_ALIGNED(struct irb, cio_irb); ^ ../include/linux/percpu-defs.h:150:2: note: expanded from macro 'DEFINE_PER_CPU_ALIGNED' DEFINE_PER_CPU_SECTION(type, name, PER_CPU_ALIGNED_SECTION) \ ^ ../include/linux/percpu-defs.h:93:9: note: expanded from macro 'DEFINE_PER_CPU_SECTION' extern __PCPU_ATTRS(sec) __typeof__(type) name; \ ^ ../include/linux/percpu-defs.h:49:26: note: expanded from macro '__PCPU_ATTRS' __percpu __attribute__((section(PER_CPU_BASE_SECTION sec))) \ ^ ../drivers/s390/cio/cio.h:118:1: note: previous attribute is here DECLARE_PER_CPU(struct irb, cio_irb); ^ ../include/linux/percpu-defs.h:111:2: note: expanded from macro 'DECLARE_PER_CPU' DECLARE_PER_CPU_SECTION(type, name, "") ^ ../include/linux/percpu-defs.h:87:9: note: expanded from macro 'DECLARE_PER_CPU_SECTION' extern __PCPU_ATTRS(sec) __typeof__(type) name ^ ../include/linux/percpu-defs.h:49:26: note: expanded from macro '__PCPU_ATTRS' __percpu __attribute__((section(PER_CPU_BASE_SECTION sec))) \ ^ Use DECLARE_PER_CPU_ALIGNED() here, to make the two match. Signed-off-by:Nathan Chancellor <natechancellor@gmail.com> Signed-off-by:
Sebastian Ott <sebott@linux.ibm.com> Signed-off-by:
-
Thomas Huth authored
If CONFIG_PGSTE is not set (e.g. when compiling without KVM), GCC complains: CC arch/s390/mm/pgtable.o arch/s390/mm/pgtable.c:413:15: warning: ‘pmd_alloc_map’ defined but not used [-Wunused-function] static pmd_t *pmd_alloc_map(struct mm_struct *mm, unsigned long addr) ^~~~~~~~~~~~~ Wrap the function with "#ifdef CONFIG_PGSTE" to silence the warning. Signed-off-by:Thomas Huth <thuth@redhat.com> Reviewed-by:
David Hildenbrand <david@redhat.com> Signed-off-by:
Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by:
-
Julian Wiedmann authored
This cursor is used for debugging only. But since commit "s390/qdio: pass up count of ready-to-process SBALs" it effectively duplicates the first_to_check cursor, diverging for just a short moment when get_*_buffer_frontier() updates q->first_to_check. Signed-off-by:
-
Julian Wiedmann authored
When passing a range of ready-to-process SBALs to the upper-layer driver, use the available 'count' instead of calculating the distance between the first_to_check and first_to_kick cursors. This simplifies the logic of the queue-scan path, and opens up the possibility of scanning all 128 SBALs in one go (as determining the reported count no longer requires wrap-around safe arithmetic on the queue's cursors). Signed-off-by:
-
Julian Wiedmann authored
When qdio_{in,out}bound_q_moved() scans a queue for pending work, it currently only returns a boolean to its caller. The interface to the upper-layer-drivers (qdio_kick_handler() and qdio_get_next_buffers()) then re-calculates the number of pending SBALs from the q->first_to_check and q->first_to_kick cursors. Refactor this so that whenever get_{in,out}bound_buffer_frontier() adjusted the queue's first_to_check cursor, it also returns the corresponding count of ready-to-process SBALs (and 0 else). A subsequent patch will then make use of this additional information. Signed-off-by: -
Julian Wiedmann authored
The DSCI is a 1-byte field, placed at the start of an u32. So when printing it to a queue's debug state, limit the output to the part that's actually occupied by the DSCI. When the DSCI is set this gives us the expected output of '1', rather than the current (obscure) value of '16777216'. Suggested-by:
-
Vasily Gorbik authored
Disallow kernel command line alteration via ipl parameter block if running in protected virtualization environment. Signed-off-by:
-
Vasily Gorbik authored
Add sharing of ipl parameter block for diag 308 set/store calls to allow kvm access in protected virtualization environment. Signed-off-by:
-
Vasily Gorbik authored
The Ultravisor Call Facility (stfle bit 158) defines an API to the Ultravisor (UV calls), a mini hypervisor located at machine level. With help of the Ultravisor, KVM will be able to run "protected" VMs, special VMs whose memory and management data are unavailable to KVM. The protected VMs can also request services from the Ultravisor. The guest api consists of UV calls to share and unshare memory with the kvm hypervisor. To enable this feature support PROTECTED_VIRTUALIZATION_GUEST kconfig option has been introduced. Co-developed-by:
Janosch Frank <frankja@de.ibm.com> Signed-off-by:
-
Vasily Gorbik authored
Same as for .boot.data section make sure that .boot.preserved.data sections of vmlinux and arch/s390/compressed/vmlinux match before producing the compressed kernel image. Symbols presence, order and sizes are cross-checked. Signed-off-by:
-
Vasily Gorbik authored
.boot.preserved.data is a better fit for ipl block than .boot.data which is discarded after init. Reusing .boot.preserved.data allows to simplify code a little bit and avoid copying data from .boot.data to persistent variables. Signed-off-by:
-
Gerald Schaefer authored
Introduce .boot.preserve.data section which is similar to .boot.data and "shared" between the decompressor code and the decompressed kernel. The decompressor will store values in it, and copy over to the decompressed image before starting it. This method allows to avoid using pre-defined addresses and other hacks to pass values between those boot phases. Unlike .boot.data section .boot.preserved.data is NOT a part of init data, and hence will be preserved for the kernel life time. Signed-off-by:
Gerald Schaefer <gerald.schaefer@de.ibm.com> Signed-off-by:
-
- 29 Mar, 2019 1 commit
-
-
Thomas Richter authored
Add support for the CPU-Measurement Facility counter second version number 6. This number is used to detect some more counters in the crypto counter set and the extended counter set. Signed-off-by:
-
