For system damage machine checks or machine checks due to invalid PSW
fields the system will be stopped. In order to get an oops message out
before killing the system the machine check handler branches to
.Lmcck_panic, switches to the panic stack and then does the usual
machine check handling.
The switch to the panic stack is incomplete, the stack pointer in %r15
is replaced, but the pt_regs pointer in %r11 is not. The result is
a program check which will kill the system in a slightly different way.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The LAST_BREAK macro in entry.S uses a different instruction sequence
for CONFIG_MARCH_Z900 builds. The branch target offset to skip the
store of the last breaking event address needs to take the different
length of the code block into account.
Fixes: f8fc82b471 ("s390: move sys_call_table and last_break from thread_info to thread_struct")
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
We have the s390 specific THREAD_ORDER define and the THREAD_SIZE_ORDER
define which is also used in common code. Both have exactly the same
semantics. Therefore get rid of THREAD_ORDER and always use
THREAD_SIZE_ORDER instead.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Move the last two architecture specific fields from the thread_info
structure to the thread_struct. All that is left in thread_info is
the flags field.
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This is the s390 variant of commit 15f4eae70d ("x86: Move
thread_info into task_struct").
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Convert s390 to use a field in the struct lowcore for the CPU
preemption count. It is a bit cheaper to access a lowcore field
compared to a thread_info variable and it removes the depencency
on a task related structure.
bloat-o-meter on the vmlinux image for the default configuration
(CONFIG_PREEMPT_NONE=y) reports a small reduction in text size:
add/remove: 0/0 grow/shrink: 18/578 up/down: 228/-5448 (-5220)
A larger improvement is achieved with the default configuration
but with CONFIG_PREEMPT=y and CONFIG_DEBUG_PREEMPT=n:
add/remove: 2/6 grow/shrink: 59/4477 up/down: 1618/-228762 (-227144)
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
After linking there are several symbols for the same address that the
__switch_to symbol points to. E.g.:
000000000089b9c0 T __kprobes_text_start
000000000089b9c0 T __lock_text_end
000000000089b9c0 T __lock_text_start
000000000089b9c0 T __sched_text_end
000000000089b9c0 T __switch_to
When disassembling with "objdump -d" this results in a missing
__switch_to function. It would be named __kprobes_text_start
instead. To unconfuse objdump add a nop in front of the kprobes text
section. That way __switch_to appears again.
Obviously this solution is sort of a hack, since it also depends on
link order if this works or not. However it is the best I can come up
with for now.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove a leftover from the code that transferred a couple of TIF bits
from the previous task to the next task.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
There is a tricky interaction between the machine check handler
and the critical sections of load_fpu_regs and save_fpu_regs
functions. If the machine check interrupts one of the two
functions the critical section cleanup will complete the function
before the machine check handler s390_do_machine_check is called.
Trouble is that the machine check handler needs to validate the
floating point registers *before* and not *after* the completion
of load_fpu_regs/save_fpu_regs.
The simplest solution is to rewind the PSW to the start of the
load_fpu_regs/save_fpu_regs and retry the function after the
return from the machine check handler.
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Cc: <stable@vger.kernel.org> # 4.3+
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
commit e22cf8ca6f ("s390/cpumf: rework program parameter setting
to detect guest samples") requires guest changes to get proper
guest/host. We can do better: We can use the primary asn value,
which is set on all Linux variants to compare this with the host
pp value.
We now have the following cases:
1. Guest using PP
host sample: gpp == 0, asn == hpp --> host
guest sample: gpp != 0 --> guest
2. Guest not using PP
host sample: gpp == 0, asn == hpp --> host
guest sample: gpp == 0, asn != hpp --> guest
As soon as the host no longer sets CR4, we must back out
this heuristics - let's add a comment in switch_to.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
It does not make sense to try to relinquish the time slice with diag 0x9c
to a CPU in a state that does not allow to schedule the CPU. The scenario
where this can happen is a CPU waiting in udelay/mdelay while holding a
spin-lock.
Add a CIF bit to tag a CPU in enabled wait and use it to detect that the
yield of a CPU will not be successful and skip the diagnose call.
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
When using systemtap it was observed that our udelay implementation is
rather suboptimal if being called from a kprobe handler installed by
systemtap.
The problem observed when a kprobe was installed on lock_acquired().
When the probe was hit the kprobe handler did call udelay, which set
up an (internal) timer and reenabled interrupts (only the clock comparator
interrupt) and waited for the interrupt.
This is an optimization to avoid that the cpu is busy looping while waiting
that enough time passes. The problem is that the interrupt handler still
does call irq_enter()/irq_exit() which then again can lead to a deadlock,
since some accounting functions may take locks as well.
If one of these locks is the same, which caused lock_acquired() to be
called, we have a nice deadlock.
This patch reworks the udelay code for the interrupts disabled case to
immediately leave the low level interrupt handler when the clock
comparator interrupt happens. That way no C code is being called and the
deadlock cannot happen anymore.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The program parameter can be used to mark hardware samples with
some token. Previously, it was used to mark guest samples only.
Improve the program parameter doubleword by combining two parts,
the leftmost LPP part and the rightmost PID part. Set the PID
part for processes by using the task PID.
To distinguish host and guest samples for the kernel (PID part
is zero), the guest must always set the program paramater to a
non-zero value. Use the leftmost bit in the LPP part of the
program parameter to be able to detect guest kernel samples.
[brueckner@linux.vnet.ibm.com]: Split __LC_CURRENT and introduced
__LC_LPP. Corrected __LC_CURRENT users and adjusted assembler parts.
And updated the commit message accordingly.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Various functions in entry.S perform test-under-mask instructions
to test for particular bits in memory. Because test-under-mask uses
a mask value of one byte, the mask value and the offset into the
memory must be calculated manually. This easily introduces errors
and is hard to review and read.
Introduce the TSTMSK assembler macro to specify a mask constant and
let the macro calculate the offset and the byte mask to generate a
test-under-mask instruction. The benefit is that existing symbolic
constants can now be used for tests. Also the macro checks for
zero mask values and mask values that consist of multiple bytes.
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Previously, the init task did not have an allocated FPU save area and
saving an FPU state was not possible. Now if the vector extension is
always enabled, provide a static FPU save area to save FPU states of
vector instructions that can be executed quite early.
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
If the kernel detects that the s390 hardware supports the vector
facility, it is enabled by default at an early stage. To force
it off, use the novx kernel parameter. Note that there is a small
time window, where the vector facility is enabled before it is
forced to be off.
With enabling the vector facility by default, the FPU save and
restore functions can be improved. They do not longer require
to manage expensive control register updates to enable or disable
the vector enablement control for particular processes.
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The calculation for the SMT scaling factor for a hardware thread
which has been partially idle needs to disregard the cycles spent
by the other threads of the core while the thread is idle.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The critical section cleanup code misses to add the offset of the
thread_struct to the task address.
Therefore, if the critical section code gets executed, it may corrupt
the task struct or restore the contents of the floating point registers
from the wrong memory location.
Fixes d0164ee20d "s390/kernel: remove save_fpu_regs() parameter and use
__LC_CURRENT instead".
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Right now we use the address of the sie control block as tag for
the sampling data. This is hard to get for users. Let's just use
the PID of the cpu thread to mark the hardware samples.
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
All calls to save_fpu_regs() specify the fpu structure of the current task
pointer as parameter. The task pointer of the current task can also be
retrieved from the CPU lowcore directly. Remove the parameter definition,
load the __LC_CURRENT task pointer from the CPU lowcore, and rebase the FPU
structure onto the task structure. Apply the same approach for the
load_fpu_regs() function.
Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
If a machine checks is received while the CPU is in the kernel, only
the s390_do_machine_check function will be called. The call to
s390_handle_mcck is postponed until the CPU returns to user space.
Because of this it is safe to use the asynchronous stack for machine
checks even if the CPU is already handling an interrupt.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Reorder the instructions of UPDATE_VTIME to improve superscalar execution,
remove duplicate checks for problem-state from the asynchronous interrupt
handlers, and move the check for problem-state from the synchronous
exit path to the program check path as it is only needed for program
checks inside the kernel.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Currently there are two mechanisms to deal with cleanup work due to
interrupts. The HANDLE_SIE_INTERCEPT macro is used to undo the changes
required to enter SIE in sie64a. If the SIE instruction causes a program
check, or an asynchronous interrupt is received the HANDLE_SIE_INTERCEPT
code forwards the program execution to sie_exit.
All the other critical sections in entry.S are handled by the code in
cleanup_critical that is called by the SWITCH_ASYNC macro.
Move the sie64a function to the beginning of the critical section and
add the code from HANDLE_SIE_INTERCEPT to cleanup_critical. Add a special
case for the sie64a cleanup to the program check handler.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The HANDLE_SIE_INTERCEPT macro is used in the interrupt handlers
and the program check handler to undo a few changes done by sie64a.
Among them are guest vs host LPP, the gmap ASCE vs kernel ASCE and
the bit that indicates that SIE is currently running on the CPU.
There is a race of a voluntary SIE exit vs asynchronous interrupts.
If the CPU completed the SIE instruction and the TM instruction of
the LPP macro at the time it receives an interrupt, the interrupt
handler will run while the LPP, the ASCE and the SIE bit are still
set up for guest execution. This might result in wrong sampling data,
but it will not cause data corruption or lockups.
The critical section in sie64a needs to be enlarged to include all
instructions that undo the changes required for guest execution.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Improve the save and restore behavior of FPU register contents to use the
vector extension within the kernel.
The kernel does not use floating-point or vector registers and, therefore,
saving and restoring the FPU register contents are performed for handling
signals or switching processes only. To prepare for using vector
instructions and vector registers within the kernel, enhance the save
behavior and implement a lazy restore at return to user space from a
system call or interrupt.
To implement the lazy restore, the save_fpu_regs() sets a CPU information
flag, CIF_FPU, to indicate that the FPU registers must be restored.
Saving and setting CIF_FPU is performed in an atomic fashion to be
interrupt-safe. When the kernel wants to use the vector extension or
wants to change the FPU register state for a task during signal handling,
the save_fpu_regs() must be called first. The CIF_FPU flag is also set at
process switch. At return to user space, the FPU state is restored. In
particular, the FPU state includes the floating-point or vector register
contents, as well as, vector-enablement and floating-point control. The
FPU state restore and clearing CIF_FPU is also performed in an atomic
fashion.
For KVM, the restore of the FPU register state is performed when restoring
the general-purpose guest registers before the SIE instructions is started.
Because the path towards the SIE instruction is interruptible, the CIF_FPU
flag must be checked again right before going into SIE. If set, the guest
registers must be reloaded again by re-entering the outer SIE loop. This
is the same behavior as if the SIE critical section is interrupted.
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
git commit 0c8c0f03e3
"x86/fpu, sched: Dynamically allocate 'struct fpu'"
moved the thread_struct to the end of the task_struct.
This causes some of the offsets used in entry.S to overflow their
instruction operand field. To fix this use aghi to create a
dedicated pointer for the thread_struct.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
exit_sie_sync is used to kick CPUs out of SIE and prevent reentering at
any point in time. This is used to reload the prefix pages and to
set the IBS stuff in a way that guarantees that after this function
returns we are no longer in SIE. All current users trigger KVM requests.
The request must be set before we block the CPUs to avoid races. Let's
make this implicit by adding the request into a new function
kvm_s390_sync_requests that replaces exit_sie_sync and split out
s390_vcpu_block and s390_vcpu_unblock, that can be used to keep
CPUs out of SIE independent of requests.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Remove the 31 bit syscalls from the syscall table. This is a separate patch
just in case I screwed something up so it can be easily reverted.
However the conversion was done with a script, so everything should be ok.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Rename a couple of files to get rid of the "64" suffix.
"git blame" will still work.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Remove the 31 bit support in order to reduce maintenance cost and
effectively remove dead code. Since a couple of years there is no
distribution left that comes with a 31 bit kernel.
The 31 bit kernel also has been broken since more than a year before
anybody noticed. In addition I added a removal warning to the kernel
shown at ipl for 5 minutes: a960062e58 ("s390: add 31 bit warning
message") which let everybody know about the plan to remove 31 bit
code. We didn't get any response.
Given that the last 31 bit only machine was introduced in 1999 let's
remove the code.
Anybody with 31 bit user space code can still use the compat mode.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
To improve the output of the perf tool hide most of the symbols
from entry[64].S by using the '.L' prefix.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
was a pretty active cycle for KVM. Changes include:
- a lot of s390 changes: optimizations, support for migration,
GDB support and more
- ARM changes are pretty small: support for the PSCI 0.2 hypercall
interface on both the guest and the host (the latter acked by Catalin)
- initial POWER8 and little-endian host support
- support for running u-boot on embedded POWER targets
- pretty large changes to MIPS too, completing the userspace interface
and improving the handling of virtualized timer hardware
- for x86, a larger set of changes is scheduled for 3.17. Still,
we have a few emulator bugfixes and support for running nested
fully-virtualized Xen guests (para-virtualized Xen guests have
always worked). And some optimizations too.
The only missing architecture here is ia64. It's not a coincidence
that support for KVM on ia64 is scheduled for removal in 3.17.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)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=j5Od
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm into next
Pull KVM updates from Paolo Bonzini:
"At over 200 commits, covering almost all supported architectures, this
was a pretty active cycle for KVM. Changes include:
- a lot of s390 changes: optimizations, support for migration, GDB
support and more
- ARM changes are pretty small: support for the PSCI 0.2 hypercall
interface on both the guest and the host (the latter acked by
Catalin)
- initial POWER8 and little-endian host support
- support for running u-boot on embedded POWER targets
- pretty large changes to MIPS too, completing the userspace
interface and improving the handling of virtualized timer hardware
- for x86, a larger set of changes is scheduled for 3.17. Still, we
have a few emulator bugfixes and support for running nested
fully-virtualized Xen guests (para-virtualized Xen guests have
always worked). And some optimizations too.
The only missing architecture here is ia64. It's not a coincidence
that support for KVM on ia64 is scheduled for removal in 3.17"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (203 commits)
KVM: add missing cleanup_srcu_struct
KVM: PPC: Book3S PR: Rework SLB switching code
KVM: PPC: Book3S PR: Use SLB entry 0
KVM: PPC: Book3S HV: Fix machine check delivery to guest
KVM: PPC: Book3S HV: Work around POWER8 performance monitor bugs
KVM: PPC: Book3S HV: Make sure we don't miss dirty pages
KVM: PPC: Book3S HV: Fix dirty map for hugepages
KVM: PPC: Book3S HV: Put huge-page HPTEs in rmap chain for base address
KVM: PPC: Book3S HV: Fix check for running inside guest in global_invalidates()
KVM: PPC: Book3S: Move KVM_REG_PPC_WORT to an unused register number
KVM: PPC: Book3S: Add ONE_REG register names that were missed
KVM: PPC: Add CAP to indicate hcall fixes
KVM: PPC: MPIC: Reset IRQ source private members
KVM: PPC: Graciously fail broken LE hypercalls
PPC: ePAPR: Fix hypercall on LE guest
KVM: PPC: BOOK3S: Remove open coded make_dsisr in alignment handler
KVM: PPC: BOOK3S: Always use the saved DAR value
PPC: KVM: Make NX bit available with magic page
KVM: PPC: Disable NX for old magic page using guests
KVM: PPC: BOOK3S: HV: Add mixed page-size support for guest
...
The oi and ni instructions used in entry[64].S to set and clear bits
in the thread-flags are not guaranteed to be atomic in regard to other
CPUs. Split the TIF bits into CPU, pt_regs and thread-info specific
bits. Updates on the TIF bits are done with atomic instructions,
updates on CPU and pt_regs bits are done with non-atomic instructions.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Always switch to the kernel ASCE in switch_mm. Load the secondary
space ASCE in finish_arch_post_lock_switch after checking that
any pending page table operations have completed. The primary
ASCE is loaded in entry[64].S. With this the update_primary_asce
call can be removed from the switch_to macro and from the start
of switch_mm function. Remove the load_primary argument from
update_user_asce/clear_user_asce, rename update_user_asce to
set_user_asce and rename update_primary_asce to load_kernel_asce.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
per_perc_atmid is currently a two-byte field that combines two
fields, the PER code and the PER Addressing-and-Translation-Mode
Identification (ATMID)
Let's make them accessible indepently and also rename per_cause to
per_code.
Signed-off-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
The current uaccess code uses a page table walk in some circumstances,
e.g. in case of the in atomic futex operations or if running on old
hardware which doesn't support the mvcos instruction.
However it turned out that the page table walk code does not correctly
lock page tables when accessing page table entries.
In other words: a different cpu may invalidate a page table entry while
the current cpu inspects the pte. This may lead to random data corruption.
Adding correct locking however isn't trivial for all uaccess operations.
Especially copy_in_user() is problematic since that requires to hold at
least two locks, but must be protected against ABBA deadlock when a
different cpu also performs a copy_in_user() operation.
So the solution is a different approach where we change address spaces:
User space runs in primary address mode, or access register mode within
vdso code, like it currently already does.
The kernel usually also runs in home space mode, however when accessing
user space the kernel switches to primary or secondary address mode if
the mvcos instruction is not available or if a compare-and-swap (futex)
instruction on a user space address is performed.
KVM however is special, since that requires the kernel to run in home
address space while implicitly accessing user space with the sie
instruction.
So we end up with:
User space:
- runs in primary or access register mode
- cr1 contains the user asce
- cr7 contains the user asce
- cr13 contains the kernel asce
Kernel space:
- runs in home space mode
- cr1 contains the user or kernel asce
-> the kernel asce is loaded when a uaccess requires primary or
secondary address mode
- cr7 contains the user or kernel asce, (changed with set_fs())
- cr13 contains the kernel asce
In case of uaccess the kernel changes to:
- primary space mode in case of a uaccess (copy_to_user) and uses
e.g. the mvcp instruction to access user space. However the kernel
will stay in home space mode if the mvcos instruction is available
- secondary space mode in case of futex atomic operations, so that the
instructions come from primary address space and data from secondary
space
In case of kvm the kernel runs in home space mode, but cr1 gets switched
to contain the gmap asce before the sie instruction gets executed. When
the sie instruction is finished cr1 will be switched back to contain the
user asce.
A context switch between two processes will always load the kernel asce
for the next process in cr1. So the first exit to user space is a bit
more expensive (one extra load control register instruction) than before,
however keeps the code rather simple.
In sum this means there is no need to perform any error prone page table
walks anymore when accessing user space.
The patch seems to be rather large, however it mainly removes the
the page table walk code and restores the previously deleted "standard"
uaccess code, with a couple of changes.
The uaccess without mvcos mode can be enforced with the "uaccess_primary"
kernel parameter.
Reported-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Git commit 050eef364a "[S390] fix tlb flushing vs. concurrent
/proc accesses" introduced the attach counter to avoid using the
mm_users value to decide between IPTE for every PTE and lazy TLB
flushing with IDTE. That fixed the problem with mm_users but it
introduced another subtle race, fortunately one that is very hard
to hit.
The background is the requirement of the architecture that a valid
PTE may not be changed while it can be used concurrently by another
cpu. The decision between IPTE and lazy TLB flushing needs to be
done while the PTE is still valid. Now if the virtual cpu is
temporarily stopped after the decision to use lazy TLB flushing but
before the invalid bit of the PTE has been set, another cpu can attach
the mm, find that flush_mm is set, do the IDTE, return to userspace,
and recreate a TLB that uses the PTE in question. When the first,
stopped cpu continues it will change the PTE while it is attached on
another cpu. The first cpu will do another IDTE shortly after the
modification of the PTE which makes the race window quite short.
To fix this race the CPU that wants to attach the address space of a
user space thread needs to wait for the end of the PTE modification.
The number of concurrent TLB flushers for an mm is tracked in the
upper 16 bits of the attach_count and finish_arch_post_lock_switch
is used to wait for the end of the flush operation if required.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Git commit 616498813b "s390: system call path micro optimization"
introduced a regression in regard to system call restarting and inferior
function calls via the ptrace interface. The pointer to the system call
table needs to be loaded in sysc_sigpending if do_signal returns with
TIF_SYSCALl set after it restored a system call context.
Cc: stable@vger.kernel.org # 3.10+
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Modify the psw_idle waiting logic in entry[64].S to return with
interrupts disabled. This avoids potential issues with udelay
and interrupt loops as interrupts are not reenabled after
clock comparator interrupts.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
With the introduction of PCI it became apparent that s390 should
convert to generic hardirqs as too many drivers do not have the
correct dependency for GENERIC_HARDIRQS. On the architecture
level s390 does not have irq lines. It has external interrupts,
I/O interrupts and adapter interrupts. This patch hard-codes all
external interrupts as irq #1, all I/O interrupts as irq #2 and
all adapter interrupts as irq #3. The additional information from
the lowcore associated with the interrupt is stored in the
pt_regs of the interrupt frame, where the interrupt handler can
pick it up. For PCI/MSI interrupts the adapter interrupt handler
scans the relevant bit fields and calls generic_handle_irq with
the virtual irq number for the MSI interrupt.
Reviewed-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Copy the interrupt parameters from the lowcore to the pt_regs structure
in entry[64].S and reduce the arguments of the low level interrupt handler
to the pt_regs pointer only. In addition move the test-pending-interrupt
loop from do_IRQ to entry[64].S to make sure that interrupt information
is always delivered via pt_regs.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Add a pointer to the system call table to the thread_info structure.
The TIF_31BIT bit is set or cleared by SET_PERSONALITY exactly once
for the lifetime of a process. With the pointer to the correct system
call table in thread_info the system call code in entry64.S path can
drop the check for TIF_31BIT which saves a couple of instructions.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Store the stack pointers in the lowcore for the kernel stack, the async
stack and the panic stack with the offset required for the first user.
This avoids an unnecessary add instruction on the system call path.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The current machine check code uses the registers stored by the machine
in the lowcore at __LC_GPREGS_SAVE_AREA as the registers of the interrupted
context. The registers 0-7 of a user process can get clobbered if a machine
checks interrupts the execution of a critical section in entry[64].S.
The reason is that the critical section cleanup code may need to modify
the PSW and the registers for the previous context to get to the end of a
critical section. If registers 0-7 have to be replaced the relevant copy
will be in the registers, which invalidates the copy in the lowcore. The
machine check handler needs to explicitly store registers 0-7 to the stack.
Cc: stable@vger.kernel.org
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Pull s390 update from Martin Schwidefsky:
"Add support to generate code for the latest machine zEC12, MOD and XOR
instruction support for the BPF jit compiler, the dasd safe offline
feature and the big one: the s390 architecture gets PCI support!!
Right before the world ends on the 21st ;-)"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (41 commits)
s390/qdio: rename the misleading PCI flag of qdio devices
s390/pci: remove obsolete email addresses
s390/pci: speed up __iowrite64_copy by using pci store block insn
s390/pci: enable NEED_DMA_MAP_STATE
s390/pci: no msleep in potential IRQ context
s390/pci: fix potential NULL pointer dereference in dma_free_seg_table()
s390/pci: use kmem_cache_zalloc instead of kmem_cache_alloc/memset
s390/bpf,jit: add support for XOR instruction
s390/bpf,jit: add support MOD instruction
s390/cio: fix pgid reserved check
vga: compile fix, disable vga for s390
s390/pci: add PCI Kconfig options
s390/pci: s390 specific PCI sysfs attributes
s390/pci: PCI hotplug support via SCLP
s390/pci: CHSC PCI support for error and availability events
s390/pci: DMA support
s390/pci: PCI adapter interrupts for MSI/MSI-X
s390/bitops: find leftmost bit instruction support
s390/pci: CLP interface
s390/pci: base support
...
The current single step code is racy in regard to concurrent delivery
of signals. If a signal is delivered after a PER program check occurred
but before the TIF_PER_TRAP bit has been checked in entry[64].S the code
clears TIF_PER_TRAP and then calls do_signal. This is wrong, if the
instruction completed (or has been suppressed) a SIGTRAP should be
delivered to the debugger in any case. Only if the instruction has been
nullified the SIGTRAP may not be send.
The new logic always sets TIF_PER_TRAP if the program check indicates PER
tracing but removes it again for all program checks that are nullifying.
The effect is that for each change in the PSW address we now get a
single SIGTRAP.
Reported-by: Andreas Arnez <arnez@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
