The current qemu hppa emulation emulates a PA1.1 CPU, which can only execute
the 32-bit instruction set. For unknown 64-bit instructions, a instruction trap
is sent to the virtual CPU.
This behaviour is correct in the sense that we emulate what the PA1.1
specification says.
But when trying to boot older Linux installation images, e.g.
ftp://parisc.parisc-linux.org/debian-cd/debian-5.0/lenny-5.0.10-hppa-iso-cd/cdimage.debian.org/debian-5010-hppa-netinst.iso
one finds that qemu fails to boot those images.
The problem is, that in the Linux kernel (e.g. 2.6.26) of those old images
64-bit instructions were used by mistake in the fault handlers. The relevant
instructions (the ",*" indicates that it's a 64-bit instruction) I see are:
0: 09 3e 04 29 sub,* sp,r9,r9
0: 08 3d 06 3d add,* ret1,r1,ret1
0: 0a 09 02 61 or,* r9,r16,r1
0: 0a ba 00 3a andcm,* r26,r21,r26
0: 08 33 02 33 and,* r19,r1,r19
The interesting part is, that real physical 32-bit machines (like the 700/64
and B160L - which is the one we emulate) do boot those images and thus seem to
simply ignore the 64-bit flag on those instructions.
The patch below modifies the qemu instruction decoder to ignore the 64-bit flag
too - which is what real 32-bit hardware seems to do. With this modification
qemu now successfully boots those older images too.
I suggest to apply the patch below - even if it does not reflect what the SPEC
says. Instead it increases the compatibility to really existing hardware and
seem to not create problems if we add real PA2.0 support anytime later.
Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Emulate a power button switch, tell SeaBIOS the address via fw_cfg and
bind the power button to the qemu UI.
Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Change QEMU_FW_CFG_IO_BASE to shorter variant FW_CFG_IO_BASE and hand
over the actual port address in %r19 to SeaBIOS.
Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Devices on hppa occupy at least 4k starting at the HPA, so MEMORY_HPA+4k is
blocked (by Linux) for the memory module. I noticed this when testing the new
Linux kernel patch to let the fw_cfg entries show up in Linux under /proc.
The Linux kernel driver could not allocate the region for fw_cfg.
This new base address seems to not conflict.
Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Until now the TLB size was fixed at 256 entries. To allow operating
systems to utilize more TLB entries in the future, we need to tell
firmware how many TLB entries we actually support in the emulation.
Firmware then reports this to the operating system via the
PDC_CACHE_INFO call.
This patch simply does the preparation to allow more TLB entries.
Signed-off-by: Helge Deller <deller@gmx.de>
We do have a QCryptoTLSCipherSuites struct. It must be used when
setting instance_size of the QOM type. Luckily this never caused
problems because the QCryptoTLSCipherSuites struct has only a
parent_obj field and nothing else.
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-Id: <20200826171005.4055015-5-ehabkost@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Currently we have a RXCPU typedef and a RXCPU type checking
macro, but OBJECT_DECLARE* would transform the RXCPU macro into a
function, and the function name would conflict with the typedef
name.
Rename the RXCPU* QOM type check macros to RX_CPU*, so we will
avoid the conflict and make the macro names consistent with the
TYPE_RX_CPU constant name.
This will make future conversion to OBJECT_DECLARE* easier.
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20200825192110.3528606-53-ehabkost@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Some QOM macros were using a X86_IOMMU_DEVICE prefix, and others
were using a X86_IOMMU prefix. Rename all of them to use the
same X86_IOMMU_DEVICE prefix.
This will make future conversion to OBJECT_DECLARE* easier.
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Message-Id: <20200825192110.3528606-47-ehabkost@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Since nvme_map_prp always operate on the request-scoped qsg/iovs, just
pass a single pointer to the NvmeRequest instead of two for each of the
qsg and iov.
Suggested-by: Minwoo Im <minwoo.im.dev@gmail.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Since clean up of the request qsg/iov is now always done post-use, there
is no need to use a stack-allocated qsg/iov in nvme_dma_prp.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Acked-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Instead of passing around the NvmeNamespace and the NvmeCmd, add them as
members in the NvmeRequest structure.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Add 'mdts' device parameter to control the Maximum Data Transfer Size of
the controller and check that it is respected.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Before this patch the device already supported PRP lists in the CMB, but
it did not check for the validity of it nor announced the support in the
Identify Controller data structure LISTS field.
If some of the PRPs in a PRP list are in the CMB, then ALL entries must
be there. This patch makes sure that requirement is verified as well as
properly announcing support for PRP lists in the CMB.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Refactor the nvme_dma_{read,write}_prp functions into a common function
taking a DMADirection parameter.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
The QSG isn't always initialized, so accounting could be wrong. Issue a
call to blk_acct_start instead with the size taken from the QSG or IOV
depending on the kind of I/O.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Add nvme_map_addr, nvme_map_addr_cmb and nvme_addr_to_cmb helpers and
use them in nvme_map_prp.
This fixes a bug where in the case of a CMB transfer, the device would
map to the buffer with a wrong length.
Fixes: b2b2b67a00 ("nvme: Add support for Read Data and Write Data in CMBs.")
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Reviewed-by: Andrzej Jakowski <andrzej.jakowski@linux.intel.com>
This is preparatory to subsequent patches that change how QSGs/IOVs are
handled. It is important that the qsg and iov members of the NvmeRequest
are initially zeroed.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Minwoo Im <minwoo.im.dev@gmail.com>
Since we are not providing the NGUID or EUI64 fields, we must support
the Namespace UUID. We do not have any way of storing a persistent
unique identifier, so conjure up a UUID that is just the namespace id.
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>
Reviewed-by: Dmitry Fomichev <dmitry.fomichev@wdc.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Message-Id: <20200706061303.246057-15-its@irrelevant.dk>
Helge Deller