The kdump-zlib data pages are not dumped from aarch64 host when the
'pvtime' is involved, that is, when the block->target_end is not aligned to
page_size. In the below example, it is expected to dump two blocks.

(qemu) info mtree -f
... ...
  00000000090a0000-00000000090a0fff (prio 0, ram): pvtime KVM
... ...
  0000000040000000-00000001bfffffff (prio 0, ram): mach-virt.ram KVM
... ...

However, there is an issue with get_next_page() so that the pages for
"mach-virt.ram" will not be dumped.

At line 1296, although we have reached at the end of the 'pvtime' block,
since it is not aligned to the page_size (e.g., 0x10000), it will not break
at line 1298.

1255 static bool get_next_page(GuestPhysBlock **blockptr, uint64_t *pfnptr,
1256                           uint8_t **bufptr, DumpState *s)
... ...
1294             memcpy(buf + addr % page_size, hbuf, n);
1295             addr += n;
1296             if (addr % page_size == 0) {
1297                 /* we filled up the page */
1298                 break;
1299             }

As a result, get_next_page() will continue to the next
block ("mach-virt.ram"). Finally, when get_next_page() returns to the
caller:

- 'pfnptr' is referring to the 'pvtime'
- but 'blockptr' is referring to the "mach-virt.ram"

When get_next_page() is called the next time, "*pfnptr += 1" still refers
to the prior 'pvtime'. It will exit immediately because it is out of the
range of the current "mach-virt.ram".

The fix is to break when it is time to come to the next block, so that both
'pfnptr' and 'blockptr' refer to the same block.

Fixes: 94d78840 ("dump: fix kdump to work over non-aligned blocks")
Cc: Joe Jin <joe.jin@oracle.com>
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20230713055819.30497-1-dongli.zhang@oracle.com>

We use the user_ss[] array to hold the user emulation sources,
and the softmmu_ss[] array to hold the system emulation ones.
Hold the latter in the 'system_ss[]' array for parity with user
emulation.

Mechanical change doing:

  $ sed -i -e s/softmmu_ss/system_ss/g $(git grep -l softmmu_ss)

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230613133347.82210-10-philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Since we *might* have user emulation with softmmu,
use the clearer 'CONFIG_SYSTEM_ONLY' key to check
for system emulation.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230613133347.82210-9-philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

This had been pulled in via qemu/plugin.h from hw/core/cpu.h,
but that will be removed.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230310195252.210956-5-richard.henderson@linaro.org>
[AJB: add various additional cases shown by CI]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20230315174331.2959-15-alex.bennee@linaro.org>
Reviewed-by: Emilio Cota <cota@braap.org>

Implement the non-x86 create_win_dump(). We can remove
the last TARGET_X86_64 #ifdef'ry in dump.c, which thus
becomes target-independent. Update meson accordingly.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-6-philmd@linaro.org>

To make dump.c less target dependent, move the TARGET_X86_64 #ifdef'ry
from dump.c to win_dump.c (introducing a win_dump_available() method
there). By doing so we can build win_dump.c on any target, and
simplify the meson rule.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-5-philmd@linaro.org>

"qemu/win_dump_defs.h" is only required by win_dump.c,
but win_dump.h requires "sysemu/dump.h" which declares
the DumpState type. Remove various unused headers.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-4-philmd@linaro.org>

TARGET_PAGE_SIZE is target specific. In preparation of
making dump.c target-agnostic, replace the compile-time
TARGET_PAGE_SIZE definition by runtime qemu_target_page_size().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-3-philmd@linaro.org>

All uses of tswap in that file are wrong, and should be using
cpu_to_dumpN, which correctly tests the endianness of the output.

Reported-by: Richard Henderson <richard.henderson@linaro.org>
Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230225094903.53167-2-philmd@linaro.org>

The only way cpu_get_note_size() can return a negative value is
integer overflow in the non-stub versions, which is a programming
error.  The stub version is not actually reachable, because the
cpu_get_dump_info() stub will fail first.  Use assert().  This gets
rid of another use of QERR_UNSUPPORTED.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20230207075115.1525-4-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>

The QERR_ macros are leftovers from the days of "rich" error objects.
We've been trying to reduce their remaining use.

Get rid of a use of QERR_UNSUPPORTED, and improve the rather vague
error message

    (qemu) dump-guest-memory mumble
    Error: this feature or command is not currently supported

to

    Error: dumping guest memory is not supported on this target

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20230207075115.1525-3-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Juan Quintela <quintela@redhat.com>
[Error message tweaked]

The has_FOO for pointer-valued FOO are redundant, except for arrays.
They are also a nuisance to work with.  Recent commit "qapi: Start to
elide redundant has_FOO in generated C" provided the means to elide
them step by step.  This is the step for qapi/dump.json.

Said commit explains the transformation in more detail.  The invariant
violations mentioned there do not occur here.

Cc: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20221104160712.3005652-14-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

include/qapi/error.h on ERRP_GUARD():

 * It must be used when the function dereferences @errp or passes
 * @errp to error_prepend(), error_vprepend(), or error_append_hint().
 * It is safe to use even when it's not needed, but please avoid
 * cluttering the source with useless code.

Clean up some of this clutter.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20221121085054.683122-3-armbru@redhat.com>

When number of CPUs utilized by guest Windows is less than defined in
QEMU (i.e., desktop versions of Windows severely limits number of CPU
sockets), patch_and_save_context routine accesses non-existent PRCB and
fails. So, limit number of processed PRCBs by NumberProcessors taken
from guest Windows driver.

Signed-off-by: Viktor Prutyanov <viktor.prutyanov@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221019235948.656411-1-viktor.prutyanov@redhat.com>

Sometimes dumping a guest from the outside is the only way to get the
data that is needed. This can be the case if a dumping mechanism like
KDUMP hasn't been configured or data needs to be fetched at a specific
point. Dumping a protected guest from the outside without help from
fw/hw doesn't yield sufficient data to be useful. Hence we now
introduce PV dump support.

The PV dump support works by integrating the firmware into the dump
process. New Ultravisor calls are used to initiate the dump process,
dump cpu data, dump memory state and lastly complete the dump process.
The UV calls are exposed by KVM via the new KVM_PV_DUMP command and
its subcommands. The guest's data is fully encrypted and can only be
decrypted by the entity that owns the customer communication key for
the dumped guest. Also dumping needs to be allowed via a flag in the
SE header.

On the QEMU side of things we store the PV dump data in the newly
introduced architecture ELF sections (storage state and completion
data) and the cpu notes (for cpu dump data).

Users can use the zgetdump tool to convert the encrypted QEMU dump to an
unencrypted one.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Message-Id: <20221017083822.43118-11-frankja@linux.ibm.com>

Add hooks which architectures can use to add arbitrary data to custom
sections.

Also add a section name string table in order to identify section
contents

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221017113210.41674-1-frankja@linux.ibm.com>

section_offset will later be used to store the offset to the section
data which will be stored last. For now memory_offset is only needed
to make section_offset look nicer.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221017083822.43118-5-frankja@linux.ibm.com>

Let's start bundling the writes of the headers and of the data so we
have a clear ordering between them. Since the ELF header uses offsets
to the headers we can freely order them.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221017083822.43118-3-frankja@linux.ibm.com>

Currently we're writing the NULL section header if we overflow the
physical header number in the ELF header. But in the future we'll add
custom section headers AND section data.

To facilitate this we need to rearange section handling a bit. As with
the other ELF headers we split the code into a prepare and a write
step.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20221017083822.43118-2-frankja@linux.ibm.com>

Rewrite get_next_page() to work over non-aligned blocks. When it
encounters non aligned addresses, it will try to fill a page provided by
the caller.

This solves a kdump crash with "tpm-crb-cmd" RAM memory region,
qemu-kvm: ../dump/dump.c:1162: _Bool get_next_page(GuestPhysBlock **,
uint64_t *, uint8_t **, DumpState *): Assertion `(block->target_start &
~target_page_mask) == 0' failed.

because:
guest_phys_block_add_section: target_start=00000000fed40080 target_end=00000000fed41000: added (count: 4)

Fixes:
https://bugzilla.redhat.com/show_bug.cgi?id=2120480



Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Acked-by: David Hildenbrand <david@redhat.com>

This should be functionally equivalent, but slightly easier to read,
with simplified paths and checks at the end of the function.

The following patch is a major rewrite to get rid of the assert().

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>

The functions in question do not actually write to the file descriptor
they set up a buffer which is later written to the fd.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220811121111.9878-9-frankja@linux.ibm.com>

Let's split the write from the modification of the elf header so we
can consolidate the write of the data in one function.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220811121111.9878-8-frankja@linux.ibm.com>

dump_calculate_size() sums up all the sizes of the guest memory
blocks. Since we already have a function that calculates the size of a
single memory block (dump_get_memblock_size()) we can simply iterate
over the blocks and use the function instead of calculating the size
ourselves.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Message-Id: <20220811121111.9878-7-frankja@linux.ibm.com>

While the DumpState begin and length variables directly mirror the API
variable names they are not very descriptive. So let's add a
"filter_area_" prefix and make has_filter a function checking length > 0.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220811121111.9878-6-frankja@linux.ibm.com>

get_start_block() returns the start address of the first memory block
or -1.

With the GuestPhysBlock iterator conversion we don't need to set the
start address and can therefore remove that code and the "start"
DumpState struct member. The only functionality left is the validation
of the start block so it only makes sense to re-name the function to
validate_start_block()

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Message-Id: <20220811121111.9878-5-frankja@linux.ibm.com>

The iteration over the memblocks in dump_iterate() is hard to
understand so it's about time to clean it up. Instead of manually
grabbing the next memblock we can use QTAILQ_FOREACH to iterate over
all memblocks.

Additionally we move the calculation of the offset and length out by
introducing and using the dump_filter_memblock_*() functions. These
functions will later be used to cleanup other parts of dump.c.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220811121111.9878-4-frankja@linux.ibm.com>

Let's make it a bit clearer that we write the program headers of the
PT_LOAD type.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Reviewed-by: Steffen Eiden <seiden@ibm.linux.com>
Message-Id: <20220811121111.9878-3-frankja@linux.ibm.com>

Before this patch, 'dump-guest-memory -w' was accepting only 64-bit
dump header provided by guest through vmcoreinfo and thus was unable
to produce 32-bit guest Windows dump. So, add 32-bit guest Windows
dumping support.

Signed-off-by: Viktor Prutyanov <viktor.prutyanov@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
[ misc error handling fixes to avoid compiler warning ]
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220406171558.199263-5-viktor.prutyanov@redhat.com>

Perform read access to Windows dump header fields via helper macros.
This is preparation for the next 32-bit guest Windows dump support.

Signed-off-by: Viktor Prutyanov <viktor.prutyanov@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220406171558.199263-3-viktor.prutyanov@redhat.com>

Context structure in 64-bit Windows differs from 32-bit one and it
should be reflected in its name.

Signed-off-by: Viktor Prutyanov <viktor.prutyanov@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220406171558.199263-2-viktor.prutyanov@redhat.com>

Just like with the other write functions let's move the 32/64 bit elf
handling to a function to improve readability.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-10-frankja@linux.ibm.com>

There's no need to have a gigantic if in there let's move the elf
32/64 bit logic into the section, segment or note code.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-9-frankja@linux.ibm.com>

There's no need to have two write functions. Let's rather have two
functions that set the data for elf 32/64 and then write it in a
common function.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-8-frankja@linux.ibm.com>

Checking d_class in dump_info leads to lengthy conditionals so let's
shorten things a bit by introducing a helper function.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-7-frankja@linux.ibm.com>

Offset calculations are easy enough to get wrong. Let's add a few
variables to make moving around elf headers and data sections easier.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20220330123603.107120-6-frankja@linux.ibm.com>

When s->shdr_num is 0 we'll add 0 bytes of section headers which is
equivalent to not adding section headers but with the multiplication
we can remove a if/else.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-5-frankja@linux.ibm.com>

Let's move from a boolean to a int variable which will later enable us
to store the number of sections that are in the dump file.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-4-frankja@linux.ibm.com>

There's no need to have phdr_num and sh_info at the same time. We can
make phdr_num 32 bit and set PN_XNUM when we write the header if
phdr_num >= PN_XNUM.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220407094824.5074-1-frankja@linux.ibm.com>

Let's move to the new way of handling errors before changing the dump
code. This patch has mostly been generated by the coccinelle script
scripts/coccinelle/errp-guard.cocci.

Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20220330123603.107120-2-frankja@linux.ibm.com>