If we encounter a new mapping while the number of available DMA entries
in vfio is 0, we are currently skipping that mapping which is a problem
if we manage to free up DMA space after that within the same RPCIT --
we will return to the guest with CC0 and have not mapped everything
within the specified range. This issue was uncovered while testing
changes to the s390 linux kernel iommu/dma code, where a different
usage pattern was employed (new mappings start at the end of the
aperture and work back towards the front, making us far more likely
to encounter new mappings before invalidated mappings during a
global refresh).
Fix this by tracking whether any mappings were skipped due to vfio
DMA limit hitting 0; when this occurs, we still continue the range
and unmap/map anything we can - then we must re-run the range again
to pickup anything that was missed. This must occur in a loop until
all requests are satisfied (success) or we detect that we are still
unable to complete all mappings (return ZPCI_RPCIT_ST_INSUFF_RES).
Link: https://lore.kernel.org/linux-s390/20221019144435.369902-1-schnelle@linux.ibm.com/
Fixes: 37fa32de70 ("s390x/pci: Honor DMA limits set by vfio")
Reported-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Matthew Rosato <mjrosato@linux.ibm.com>
Message-Id: <20221028194758.204007-2-mjrosato@linux.ibm.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Revert the control and flag bits in the subchannel status word in case
the SSCH operation fails with non-zero CC (ditto for CSCH and HSCH).
According to POPS, the control and flag bits are only changed if SSCH,
CSCH, and HSCH return CC 0, and no other action should be taken otherwise.
In order to simulate that after the fact, the bits need to be reverted on
non-zero CC.
While the do_subchannel_work logic for virtual (virtio) devices will
return condition code 0, passthrough (vfio) devices may encounter
errors from either the host kernel or real hardware that need to be
accounted for after this point. This includes restoring the state of
the Subchannel Status Word to reflect the subchannel, as these bits
would not be set in the event of a non-zero condition code from the
affected instructions.
Experimentation has shown that a failure on a START SUBCHANNEL (SSCH)
to a passthrough device would leave the subchannel with the START
PENDING activity control bit set, thus blocking subsequent SSCH
operations in css_do_ssch() until some form of error recovery was
undertaken since no interrupt would be expected.
Signed-off-by: Peter Jin <pjin@linux.ibm.com>
Message-Id: <20221027212341.2904795-1-pjin@linux.ibm.com>
Reviewed-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
[thuth: Updated the commit description to Eric's suggestion]
Signed-off-by: Thomas Huth <thuth@redhat.com>
The clang-user test exceeds the 1 hour timeout occassionally.
Philippe Mathieu-Daudé has pointed out that the number of tcg tests has
increased since QEMU 7.1. The execution time therefore probably reflects
a legitimate increase in tests rather than a performance regression.
Bump the timeout to prevent CI failures.
Suggested-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20221104113659.427690-1-stefanha@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Nemonics E1000_STATUS_LAN_INIT_DONE and E1000_STATUS_ASDV_1000 have
the same value, and E1000_STATUS_ASDV_1000 should be used here because
E1000_STATUS_ASDV_1000 represents the auto-detected speed tested here
while E1000_STATUS_LAN_INIT_DONE is a value used for a different purpose
with a variant of e1000e family different from the one implemented in
QEMU.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20221103083425.100590-1-akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
The register definitions in tests/qtest/e1000e-test.c had names
different from hw/net/e1000_regs.h, which made it hard to understand
what test codes corresponds to the implementation. Use
hw/net/e1000_regs.h from tests/qtest/libqos/e1000e.c to remove
these duplications.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20221103095416.110162-1-akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
The later device status check depends on E1000_STATUS_LU, which is
enabled by E1000_CTRL_SLU. Though E1000_STATUS_LU is not implemented
and E1000_STATUS_LU is always available in the current implementation,
be a bit nicer and set E1000_CTRL_SLU just in case the bit is
implemented in the future.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20221103025451.27446-1-akihiko.odaki@daynix.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
if QEMU is configured with modules enabled, it is possible that the
load of an accelerator module will fail.
Exit in this case, relying on module_object_class_by_name to report
the specific load error if any.
Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
[claudio: changed abort() to exit(1)]
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20220929093035.4231-6-cfontana@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
improve error handling during module load, by changing:
bool module_load(const char *prefix, const char *lib_name);
void module_load_qom(const char *type);
to:
int module_load(const char *prefix, const char *name, Error **errp);
int module_load_qom(const char *type, Error **errp);
where the return value is:
-1 on module load error, and errp is set with the error
0 on module or one of its dependencies are not installed
1 on module load success
2 on module load success (module already loaded or built-in)
module_load_qom_one has been introduced in:
commit 28457744c3 ("module: qom module support"), which built on top of
module_load_one, but discarded the bool return value. Restore it.
Adapt all callers to emit errors, or ignore them, or fail hard,
as appropriate in each context.
Replace the previous emission of errors via fprintf in _some_ error
conditions with Error and error_report, so as to emit to the appropriate
target.
A memory leak is also fixed as part of the module_load changes.
audio: when attempting to load an audio module, report module load errors.
Note that still for some callers, a single issue may generate multiple
error reports, and this could be improved further.
Regarding the audio code itself, audio_add() seems to ignore errors,
and this should probably be improved.
block: when attempting to load a block module, report module load errors.
For the code paths that already use the Error API, take advantage of those
to report module load errors into the Error parameter.
For the other code paths, we currently emit the error, but this could be
improved further by adding Error parameters to all possible code paths.
console: when attempting to load a display module, report module load errors.
qdev: when creating a new qdev Device object (DeviceState), report load errors.
If a module cannot be loaded to create that device, now abort execution
(if no CONFIG_MODULE) or exit (if CONFIG_MODULE).
qom/object.c: when initializing a QOM object, or looking up class_by_name,
report module load errors.
qtest: when processing the "module_load" qtest command, report errors
in the load of the module.
Signed-off-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220929093035.4231-4-cfontana@suse.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This fixes some compiler warnings with compiler flag
-Wmissing-variable-declarations (tested with clang):
aarch64_be-linux-user-gdbstub-xml.c:564:19: warning: no previous extern declaration for non-static variable 'xml_builtin' [-Wmissing-variable-declarations]
aarch64-linux-user-gdbstub-xml.c:564:19: warning: no previous extern declaration for non-static variable 'xml_builtin' [-Wmissing-variable-declarations]
aarch64-softmmu-gdbstub-xml.c:1763:19: warning: no previous extern declaration for non-static variable 'xml_builtin' [-Wmissing-variable-declarations]
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
meson.build has one test where "main" is declared unnecessarily
with argc and argv arguments, but does not use them. Because
the test needs -Werror too, HAVE_BROKEN_SIZE_MAX is defined
incorrectly.
Fix the test and, for consistency, remove argc and argv whenever
they are not needed.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The configure script fails because it tries to compile small C programs
with a main function which is declared with arguments argc and argv
although those arguments are unused.
Running `configure -extra-cflags=-Wunused-parameter` triggers the problem.
configure for a native build does abort but shows the error in config.log.
A cross build configure for Windows with Debian stable aborts with an
error.
Avoiding unused arguments fixes this.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
Message-Id: <20221102202258.456359-1-sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Before this commit, there were contradictory descriptions about size of EFER
register.
Line 113 says the size is 8 bytes.
Line 129 says the size is 4 bytes.
As a result, when GDB is debugging an OS running on QEMU, the GDB cannot
read 'g' packets correctly. This 'g' packet transmits values of each
registers of machine emulated by QEMU to GDB. QEMU, the packet sender,
assign 4 bytes for EFER in 'g' packet based on the line 113.
GDB, the packet receiver, extract 8 bytes for EFER in 'g' packet based on
the line 129. Therefore, all registers located behind EFER in 'g' packet
has been shifted 4 bytes in GDB.
After this commit, GDB can read 'g' packets correctly.
Signed-off-by: TaiseiIto <taisei1212@outlook.jp>
Message-Id: <TY0PR0101MB4285F637209075C9F65FCDA6A4479@TY0PR0101MB4285.apcprd01.prod.exchangelabs.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Fix the logic in qemu_add_wait_object() to avoid adding the same
HANDLE twice, as the behavior is undefined when passing an array
that contains same HANDLEs to WaitForMultipleObjects() API.
Signed-off-by: Bin Meng <bin.meng@windriver.com>
Message-Id: <20221019102015.2441622-2-bmeng.cn@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
We use the .h.inc extension to include C headers. To be consistent
with the rest of the codebase, rename the C headers using the .def
extension.
IDE/tools using our .editorconfig / .gitattributes will leverage
this consistency.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20221025235006.7215-4-philmd@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
We use the .h.inc extension to include C headers. To be consistent
with the rest of the codebase, rename the C headers using the .def
extension.
IDE/tools using our .editorconfig / .gitattributes will leverage
this consistency.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20221025235006.7215-3-philmd@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
We use the .h.inc extension to include C headers. To be consistent
with the rest of the codebase, rename the C headers using the .def
extension.
IDE/tools using our .editorconfig / .gitattributes will leverage
this consistency.
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20221025235006.7215-2-philmd@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
We need to check HCR_E2H and HCR_TGE to select the right MMU index for
the correct translation regime.
To check for EL2&0 translation regime:
- For S1E0*, S1E1* and S12E* ops, check both HCR_E2H and HCR_TGE
- For S1E2* ops, check only HCR_E2H
Signed-off-by: Ake Koomsin <ake@igel.co.jp>
Message-id: 20221101064250.12444-1-ake@igel.co.jp
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
When we implemented the PAN support we theoretically wanted
to support it for both AArch32 and AArch64, but in practice
several bugs made it essentially unusable with an AArch32
guest. Fix all those problems:
- Use CPSR.PAN to check for PAN state in aarch32 mode
- throw permission fault during address translation when PAN is
enabled and kernel tries to access user acessible page
- ignore SCTLR_XP bit for armv7 and armv8 (conflicts with SCTLR_SPAN).
Signed-off-by: Timofey Kutergin <tkutergin@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20221027112619.2205229-1-tkutergin@gmail.com
[PMM: tweak commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
The HCR_EL2.TTLB bit is supposed to trap all EL1 execution of TLB
maintenance instructions. However we have added new TLB insns for
FEAT_TLBIOS and FEAT_TLBIRANGE, and forgot to set their accessfn to
access_ttlb. Add the missing accessfns.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
When we direct boot a kernel on a CPU which emulates EL3, we need
to set up the EL3 system registers as the Linux kernel documentation
specifies:
https://www.kernel.org/doc/Documentation/arm64/booting.rst
For SVE and SME this includes:
- ZCR_EL3.LEN must be initialised to the same value for all CPUs the
kernel is executed on.
- SMCR_EL3.LEN must be initialised to the same value for all CPUs the
kernel will execute on.
Although we are technically compliant with this, the "same value" we
currently use by default is the reset value of 0. This will end up
forcing the guest kernel's SVE and SME vector length to be only the
smallest supported length.
Initialize the vector length fields to their maximum possible value,
which is 0xf. If the implementation doesn't actually support that
vector length then the effective vector length will be constrained
down to the maximum supported value at point of use.
This allows the guest to use all the vector lengths the emulated CPU
supports (by programming the _EL2 and _EL1 versions of these
registers.)
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20221027140207.413084-2-peter.maydell@linaro.org
When the address is invalid address, We should set exception_index
according to MMUAccessType, and EXCCODE_ADEF need't update badinstr.
Otherwise, The system enters an infinite loop. e.g:
run test.c on system mode
test.c:
#include<stdio.h>
void (*func)(int *);
int main()
{
int i = 8;
void *ptr = (void *)0x4000000000000000;
func = ptr;
func(&i);
return 0;
}
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20221101073210.3934280-2-gaosong@loongson.cn>
1. When cpu read or write extioi COREISR reg, it should access
the reg belonged to itself, so the cpu index of 's->coreisr'
is current cpu number. Using MemTxAttrs' requester_id to get
the cpu index.
2. it need not to mask 0x1f when calculate the coreisr array index.
Signed-off-by: Xiaojuan Yang <yangxiaojuan@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20221021015307.2570844-3-yangxiaojuan@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
