The recent XFER_TO_GUEST_WORK change resulted in a situation, where the
vsie code would interpret a signal during work as a machine check during
SIE as both use the EINTR return code.
The exit_reason of the sie64a function has nothing to do with the
kvm_run exit_reason. Rename it and define a specific code for machine
checks instead of abusing -EINTR.
rename exit_reason into sie_return to avoid the naming conflict
and change the code flow in vsie.c to have a separate variable for rc
and sie_return.
Fixes: 2bd1337a12 ("KVM: s390: Use generic VIRT_XFER_TO_GUEST_WORK functions")
Signed-off-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Reviewed-by: Heiko Carstens <hca@linux.ibm.com>
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
s390_handle_damage() ends by calling the non-returning function
disabled_wait() and therefore also never returns. Annotate it with the
__noreturn compiler attribute to improve compiler optimizations.
Remove the unreachable infinite while loop.
Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
In case of an unrecoverable machine check only the machine check interrupt
code is printed to the console before the machine is stopped. This makes
root cause analysis sometimes hard.
Print additional machine check information to make analysis easier.
The output now looks like this:
Unrecoverable machine check, code: 00400F5F4C3B0000
6.16.0-rc2-11605-g987a9431e53a-dirty
HW: IBM 3931 A01 704 (z/VM 7.4.0)
PSW: 0706C00180000000 000003FFE0F0462E PFX: 0000000000070000
LBA: 000003FFE0F0462A EDC: 0000000000000000 FSA: 0000000000000000
CRS:
0080000014966A12 0000000087CB41C7 0000000000BFF140 0000000000000000
000000000000FFFF 0000000000BFF140 0000000071000000 0000000087CB41C7
0000000000008000 0000000000000000 0000000000000000 0000000000000000
0000000000000000 00000000024C0007 00000000DB000000 0000000000BFF000
GPRS:
FFFFFFFF00000000 000003FFE0F0462E E10EA4F489F897A6 0000000000000000
7FFFFFF2C0413C4C 000003FFE19B7010 0000000000000000 0000000000000000
0000000000000000 00000001F76B3380 000003FFE15D4050 0000000000000005
0000000000000000 0000000000070000 000003FFE0F0586C 0000037FE00B7DA0
System stopped
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Remove include <linux/export.h> from all files which do not contain an
EXPORT_SYMBOL().
See commit 7d95680d64 ("scripts/misc-check: check unnecessary #include
<linux/export.h> when W=1") for more details.
Acked-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Convert MACHINE_HAS_... to cpu_has_...() which uses test_facility() instead
of testing the machine_flags lowcore member if the feature is present.
test_facility() generates better code since it results in a static branch
without accessing memory. The branch is patched via alternatives by the
decompressor depending on the availability of the required facility.
Reviewed-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Assign the output from get_lowcore() to a local variable,
so the code is easier to read.
Acked-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Move, rename, and merge the fpu and vx header files. This way fpu header
files have a consistent naming scheme (fpu*.h).
Also get rid of the fpu subdirectory and move header files to asm
directory, so that all fpu and vx header files can be found at the same
location.
Merge internal.h header file into other header files, since the internal
helpers are used at many locations. so those helper functions are really
not internal.
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Remove the historic machine check handler code which validates registers.
Registers are automatically validated as part of the machine check handling
sequence (see Principles of Operation, Machine-Check Handling chapter,
Validation).
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Move the switch_to() implementation to process.c and use the generic
switch_to.h header file instead, like some other architectures.
This addresses also the oddity that the old switch_to() implementation
assigns the return value of __switch_to() to 'prev' instead of 'last',
like it should.
Remove also all includes of switch_to.h from C files, except process.c.
Acked-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Get rid of MACHINE_HAS_VX and replace it with cpu_has_vx() which is a
short readable wrapper for "test_facility(129)".
Facility bit 129 is set if the vector facility is present. test_facility()
returns also true for all bits which are set in the architecture level set
of the cpu that the kernel is compiled for. This means that
test_facility(129) is a compile time constant which returns true for z13
and later, since the vector facility bit is part of the z13 kernel ALS.
In result the compiled code will have less runtime checks, and less code.
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Instead of using local_mcck_disable() / local_mcck_enable() implement and
use local_mcck_save() / local_mcck_restore() to disable machine checks, and
restoring the previous state.
The problem with using local_mcck_disable() / local_mcck_enable() is that
there is an assumption that machine checks are always enabled. While this
is currently the case the code still looks quite odd, readers need to
double check if the code is correct.
In order to increase readability save and then restore the old machine
check mask bit, instead of assuming that it must have been enabled.
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Use control register bit defines instead of plain numbers where
possible.
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Convert all single control register usages of __local_ctl_load() and
__local_ctl_store() to local_ctl_load() and local_ctl_store().
This also requires to change the type of some struct lowcore members
from __u64 to unsigned long.
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Add local and system prefix to some functions to clarify they change
control register contents on either the local CPU or the on all CPUs.
This results in the following API:
Two defines which load and save multiple control registers.
The defines correlate with the following C prototypes:
void __local_ctl_load(unsigned long *, unsigned int cr_low, unsigned int cr_high);
void __local_ctl_store(unsigned long *, unsigned int cr_low, unsigned int cr_high);
Two functions which locally set or clear one bit for a specified
control register:
void local_ctl_set_bit(unsigned int cr, unsigned int bit);
void local_ctl_clear_bit(unsigned int cr, unsigned int bit);
Two functions which set or clear one bit for a specified control
register on all CPUs:
void system_ctl_set_bit(unsigned int cr, unsigned int bit);
void system_ctl_clear_bit(unsigend int cr, unsigned int bit);
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Rename ctl_reg.h to ctlreg.h so it matches not only ctlreg.c but also
other control register related function, union, and structure names,
which all come with a ctlreg prefix.
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
If a machine check interrupt hits while user process is
running __s390_handle_mcck() helper function is called
directly from the interrupt handler and terminates the
current process by calling make_task_dead() routine.
The make_task_dead() is not allowed to be called from
interrupt context which forces the machine check handler
switch to the kernel stack and enable local interrupts
first.
The __s390_handle_mcck() could also be called to service
pending work, but this time from the external interrupts
handler. It is the machine check handler that establishes
the work and schedules the external interrupt, therefore
the machine check interrupt itself should be disabled
while reading out the corresponding variable:
local_mcck_disable();
mcck = *this_cpu_ptr(&cpu_mcck);
memset(this_cpu_ptr(&cpu_mcck), 0, sizeof(mcck));
local_mcck_enable();
However, local_mcck_disable() does not have effect when
__s390_handle_mcck() is called directly form the machine
check handler, since the machine check interrupt is still
disabled. Therefore, it is not the opening bracket to the
following local_mcck_enable() function.
Simplify the user process termination flow by scheduling
the external interrupt and killing the affected process
from the interrupt context.
Assume a kernel-generated signal is always delivered and
ignore a value returned by do_send_sig_info() funciton.
Reviewed-by: Heiko Carstens <hca@linux.ibm.com>
Reviewed-by: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
When a machine check is received while in SIE, it is reinjected into the
guest in some cases. The respective code needs to access the sie_block,
which is taken from the backed up R14.
Since reinjection only occurs while we are in SIE (i.e. between the
labels sie_entry and sie_leave in entry.S and thus if CIF_MCCK_GUEST is
set), the backed up R14 will always contain a physical address in
s390_backup_mcck_info.
This currently works, because virtual and physical addresses are
the same.
Add phys_to_virt() to resolve the virtual-physical confusion.
Signed-off-by: Nico Boehr <nrb@linux.ibm.com>
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Link: https://lore.kernel.org/r/20230216121208.4390-2-nrb@linux.ibm.com
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Get rid of private "nmi_save_areas" slab cache. The only reason this was
introduced years ago was that with some slab debugging options allocations
would only guarantee a minimum alignment of ARCH_KMALLOC_MINALIGN, which
was eight bytes back then. This is not sufficient for the extended machine
check save area.
However since commit 59bb47985c ("mm, sl[aou]b: guarantee natural
alignment for kmalloc(power-of-two)") kmalloc guarantees a power-of-two
alignment even with debugging options enabled.
Therefore the private slab cache can be removed.
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Checking for storage errors in machine check entry code was done in order
to handle also storage errors on kernel page tables. However this is
extremely unlikely and some basic assumptions what works on machine check
entry are necessary anyway. In order to simplify machine check handling
delay checking for storage errors to C code.
With this also change the machine check new PSW to have DAT on, which
simplifies the entry code even further.
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
In case a system will be stopped because of e.g. missing validity bits
print the machine check interruption code before the system is stopped.
This is helpful, since up to now no message was printed in such a
case. Only a disabled wait PSW was loaded, which doesn't give a hint of
what went wrong.
Improve this by printing a message with debug information.
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
If a machine check happens in kernel mode, and the machine check
interruption code indicates that e.g. vector register contents in the
machine check area are not valid, the logic is to kill current.
The idea behind this was that if within kernel context vector
registers are not used then it is sufficient to kill the current user
space process to avoid that it continues with potentially corrupt
register contents. This however does not necessarily work, since the
current code does not take into account that a machine check can also
happen when a kernel thread is running (= no user space context), and
in addition there is no way to distinguish between the "previous" and
"next" user process task, if the machine check happens when a task
switch happens.
Given that machine checks with invalid saved register contents in the
machine check save area are extremely rare, simplify the logic: if
register contents are invalid and the previous context was kernel
mode, stop the whole machine. If the previous context was user mode,
kill the corresponding task.
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Use vector instruction macros instead of byte patterns to increase
readability. The generated code is nearly identical:
- 1e8: e7 0f 10 00 00 36 vlm %v0,%v15,0(%r1)
- 1ee: e7 0f 11 00 0c 36 vlm %v16,%v31,256(%r1)
+ 1e8: e7 0f 10 00 30 36 vlm %v0,%v15,0(%r1),3
+ 1ee: e7 0f 11 00 3c 36 vlm %v16,%v31,256(%r1),3
By using the VLM macro the alignment hint is automatically specified
too. Even though from a performance perspective it doesn't matter at
all for the machine check code, this shows yet another benefit when
using the macros.
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
