From: Pierrick Bouvier <pierrick.bouvier@linaro.org>
To: "Maciej S. Szmigiero" <mail@maciej.szmigiero.name>
Cc: kvm@vger.kernel.org, philmd@linaro.org, qemu-devel@nongnu.org,
Paolo Bonzini <pbonzini@redhat.com>,
manos.pitsidianakis@linaro.org, richard.henderson@linaro.org,
Marcelo Tosatti <mtosatti@redhat.com>,
alex.bennee@linaro.org
Subject: Re: [PATCH 3/7] hw/hyperv/vmbus: common compilation unit
Date: Fri, 7 Mar 2025 09:27:58 -0800 [thread overview]
Message-ID: <de815285-78a8-4b82-b6e3-e3315a25ed8f@linaro.org> (raw)
In-Reply-To: <93aa0f81-06c6-4b6b-9d8b-fcae0c17f488@maciej.szmigiero.name>
On 3/7/25 03:03, Maciej S. Szmigiero wrote:
> Hi Pierrick,
>
> On 6.03.2025 23:59, Pierrick Bouvier wrote:
>> Hi Maciej,
>>
>> we are currently working toward building a single QEMU binary able to emulate all architectures, and one prerequisite is to remove duplication of compilation units (some are duplicated per target now, because of compile time defines).
>>
>> So the work here is to replace those compile time defines with runtime functions instead, so the same code can be used for various architectures.
>
> But this is x86-only where AFAIK page size is always 4k
> so is TARGET_PAGE_SIZE going away eventually or is the
> QEMU policy to get rid of it at the first opportunity?
>
Policy is a strong word, but this is the work we are trying to do at the
moment. Please note that Richard is trying to convert those globally to
runtime checks:
https://patchew.org/QEMU/20250306234108.378881-1-richard.henderson@linaro.org/
While this is not upstream, we try to move forward with what is
available, and we can do the replacement again later.
>> Is it more clear for you?
>
> Thanks,
> Maciej
>
>> On 3/6/25 12:29, Maciej S. Szmigiero wrote:
>>> On 6.03.2025 07:41, Pierrick Bouvier wrote:
>>>> Replace TARGET_PAGE.* by runtime calls.
>>>
>>> Seems like this patch subject/title is not aligned
>>> well with its content, or a least incomplete.
>>>
>>> Also, could you provide more detailed information
>>> why TARGET_PAGE_SIZE is getting replaced by
>>> qemu_target_page_size() please?
>>>
>>> I don't see such information in the cover letter either.
>>>
>>> Thanks,
>>> Maciej
>>>> Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
>>>> ---
>>>> hw/hyperv/vmbus.c | 50 +++++++++++++++++++++----------------------
>>>> hw/hyperv/meson.build | 2 +-
>>>> 2 files changed, 26 insertions(+), 26 deletions(-)
>>>>
>>>> diff --git a/hw/hyperv/vmbus.c b/hw/hyperv/vmbus.c
>>>> index 12a7dc43128..109ac319caf 100644
>>>> --- a/hw/hyperv/vmbus.c
>>>> +++ b/hw/hyperv/vmbus.c
>>>> @@ -18,7 +18,7 @@
>>>> #include "hw/hyperv/vmbus.h"
>>>> #include "hw/hyperv/vmbus-bridge.h"
>>>> #include "hw/sysbus.h"
>>>> -#include "cpu.h"
>>>> +#include "exec/target_page.h"
>>>> #include "trace.h"
>>>> enum {
>>>> @@ -309,7 +309,7 @@ void vmbus_put_gpadl(VMBusGpadl *gpadl)
>>>> uint32_t vmbus_gpadl_len(VMBusGpadl *gpadl)
>>>> {
>>>> - return gpadl->num_gfns * TARGET_PAGE_SIZE;
>>>> + return gpadl->num_gfns * qemu_target_page_size();
>>>> }
>>>> static void gpadl_iter_init(GpadlIter *iter, VMBusGpadl *gpadl,
>>>> @@ -323,14 +323,14 @@ static void gpadl_iter_init(GpadlIter *iter, VMBusGpadl *gpadl,
>>>> static inline void gpadl_iter_cache_unmap(GpadlIter *iter)
>>>> {
>>>> - uint32_t map_start_in_page = (uintptr_t)iter->map & ~TARGET_PAGE_MASK;
>>>> - uint32_t io_end_in_page = ((iter->last_off - 1) & ~TARGET_PAGE_MASK) + 1;
>>>> + uint32_t map_start_in_page = (uintptr_t)iter->map & ~qemu_target_page_mask();
>>>> + uint32_t io_end_in_page = ((iter->last_off - 1) & ~qemu_target_page_mask()) + 1;
>>>> /* mapping is only done to do non-zero amount of i/o */
>>>> assert(iter->last_off > 0);
>>>> assert(map_start_in_page < io_end_in_page);
>>>> - dma_memory_unmap(iter->as, iter->map, TARGET_PAGE_SIZE - map_start_in_page,
>>>> + dma_memory_unmap(iter->as, iter->map, qemu_target_page_size() - map_start_in_page,
>>>> iter->dir, io_end_in_page - map_start_in_page);
>>>> }
>>>> @@ -348,17 +348,17 @@ static ssize_t gpadl_iter_io(GpadlIter *iter, void *buf, uint32_t len)
>>>> assert(iter->active);
>>>> while (len) {
>>>> - uint32_t off_in_page = iter->off & ~TARGET_PAGE_MASK;
>>>> - uint32_t pgleft = TARGET_PAGE_SIZE - off_in_page;
>>>> + uint32_t off_in_page = iter->off & ~qemu_target_page_mask();
>>>> + uint32_t pgleft = qemu_target_page_size() - off_in_page;
>>>> uint32_t cplen = MIN(pgleft, len);
>>>> void *p;
>>>> /* try to reuse the cached mapping */
>>>> if (iter->map) {
>>>> uint32_t map_start_in_page =
>>>> - (uintptr_t)iter->map & ~TARGET_PAGE_MASK;
>>>> - uint32_t off_base = iter->off & ~TARGET_PAGE_MASK;
>>>> - uint32_t mapped_base = (iter->last_off - 1) & ~TARGET_PAGE_MASK;
>>>> + (uintptr_t)iter->map & ~qemu_target_page_mask();
>>>> + uint32_t off_base = iter->off & ~qemu_target_page_mask();
>>>> + uint32_t mapped_base = (iter->last_off - 1) & ~qemu_target_page_mask();
>>>> if (off_base != mapped_base || off_in_page < map_start_in_page) {
>>>> gpadl_iter_cache_unmap(iter);
>>>> iter->map = NULL;
>>>> @@ -368,10 +368,10 @@ static ssize_t gpadl_iter_io(GpadlIter *iter, void *buf, uint32_t len)
>>>> if (!iter->map) {
>>>> dma_addr_t maddr;
>>>> dma_addr_t mlen = pgleft;
>>>> - uint32_t idx = iter->off >> TARGET_PAGE_BITS;
>>>> + uint32_t idx = iter->off >> qemu_target_page_bits();
>>>> assert(idx < iter->gpadl->num_gfns);
>>>> - maddr = (iter->gpadl->gfns[idx] << TARGET_PAGE_BITS) | off_in_page;
>>>> + maddr = (iter->gpadl->gfns[idx] << qemu_target_page_bits()) | off_in_page;
>>>> iter->map = dma_memory_map(iter->as, maddr, &mlen, iter->dir,
>>>> MEMTXATTRS_UNSPECIFIED);
>>>> @@ -382,7 +382,7 @@ static ssize_t gpadl_iter_io(GpadlIter *iter, void *buf, uint32_t len)
>>>> }
>>>> }
>>>> - p = (void *)(uintptr_t)(((uintptr_t)iter->map & TARGET_PAGE_MASK) |
>>>> + p = (void *)(uintptr_t)(((uintptr_t)iter->map & qemu_target_page_mask()) |
>>>> off_in_page);
>>>> if (iter->dir == DMA_DIRECTION_FROM_DEVICE) {
>>>> memcpy(p, buf, cplen);
>>>> @@ -591,9 +591,9 @@ static void ringbuf_init_common(VMBusRingBufCommon *ringbuf, VMBusGpadl *gpadl,
>>>> uint32_t begin, uint32_t end)
>>>> {
>>>> ringbuf->as = as;
>>>> - ringbuf->rb_addr = gpadl->gfns[begin] << TARGET_PAGE_BITS;
>>>> - ringbuf->base = (begin + 1) << TARGET_PAGE_BITS;
>>>> - ringbuf->len = (end - begin - 1) << TARGET_PAGE_BITS;
>>>> + ringbuf->rb_addr = gpadl->gfns[begin] << qemu_target_page_bits();
>>>> + ringbuf->base = (begin + 1) << qemu_target_page_bits();
>>>> + ringbuf->len = (end - begin - 1) << qemu_target_page_bits();
>>>> gpadl_iter_init(&ringbuf->iter, gpadl, as, dir);
>>>> }
>>>> @@ -734,7 +734,7 @@ static int vmbus_channel_notify_guest(VMBusChannel *chan)
>>>> unsigned long *int_map, mask;
>>>> unsigned idx;
>>>> hwaddr addr = chan->vmbus->int_page_gpa;
>>>> - hwaddr len = TARGET_PAGE_SIZE / 2, dirty = 0;
>>>> + hwaddr len = qemu_target_page_size() / 2, dirty = 0;
>>>> trace_vmbus_channel_notify_guest(chan->id);
>>>> @@ -743,7 +743,7 @@ static int vmbus_channel_notify_guest(VMBusChannel *chan)
>>>> }
>>>> int_map = cpu_physical_memory_map(addr, &len, 1);
>>>> - if (len != TARGET_PAGE_SIZE / 2) {
>>>> + if (len != qemu_target_page_size() / 2) {
>>>> res = -ENXIO;
>>>> goto unmap;
>>>> }
>>>> @@ -1038,14 +1038,14 @@ static int sgl_from_gpa_ranges(QEMUSGList *sgl, VMBusDevice *dev,
>>>> }
>>>> len -= sizeof(range);
>>>> - if (range.byte_offset & TARGET_PAGE_MASK) {
>>>> + if (range.byte_offset & qemu_target_page_mask()) {
>>>> goto eio;
>>>> }
>>>> for (; range.byte_count; range.byte_offset = 0) {
>>>> uint64_t paddr;
>>>> uint32_t plen = MIN(range.byte_count,
>>>> - TARGET_PAGE_SIZE - range.byte_offset);
>>>> + qemu_target_page_size() - range.byte_offset);
>>>> if (len < sizeof(uint64_t)) {
>>>> goto eio;
>>>> @@ -1055,7 +1055,7 @@ static int sgl_from_gpa_ranges(QEMUSGList *sgl, VMBusDevice *dev,
>>>> goto err;
>>>> }
>>>> len -= sizeof(uint64_t);
>>>> - paddr <<= TARGET_PAGE_BITS;
>>>> + paddr <<= qemu_target_page_bits();
>>>> paddr |= range.byte_offset;
>>>> range.byte_count -= plen;
>>>> @@ -1804,7 +1804,7 @@ static void handle_gpadl_header(VMBus *vmbus, vmbus_message_gpadl_header *msg,
>>>> * anything else and simplify things greatly.
>>>> */
>>>> if (msg->rangecount != 1 || msg->range[0].byte_offset ||
>>>> - (msg->range[0].byte_count != (num_gfns << TARGET_PAGE_BITS))) {
>>>> + (msg->range[0].byte_count != (num_gfns << qemu_target_page_bits()))) {
>>>> return;
>>>> }
>>>> @@ -2240,10 +2240,10 @@ static void vmbus_signal_event(EventNotifier *e)
>>>> return;
>>>> }
>>>> - addr = vmbus->int_page_gpa + TARGET_PAGE_SIZE / 2;
>>>> - len = TARGET_PAGE_SIZE / 2;
>>>> + addr = vmbus->int_page_gpa + qemu_target_page_size() / 2;
>>>> + len = qemu_target_page_size() / 2;
>>>> int_map = cpu_physical_memory_map(addr, &len, 1);
>>>> - if (len != TARGET_PAGE_SIZE / 2) {
>>>> + if (len != qemu_target_page_size() / 2) {
>>>> goto unmap;
>>>> }
>>>> diff --git a/hw/hyperv/meson.build b/hw/hyperv/meson.build
>>>> index f4aa0a5ada9..c855fdcf04c 100644
>>>> --- a/hw/hyperv/meson.build
>>>> +++ b/hw/hyperv/meson.build
>>>> @@ -1,6 +1,6 @@
>>>> specific_ss.add(when: 'CONFIG_HYPERV', if_true: files('hyperv.c'))
>>>> specific_ss.add(when: 'CONFIG_HYPERV_TESTDEV', if_true: files('hyperv_testdev.c'))
>>>> -specific_ss.add(when: 'CONFIG_VMBUS', if_true: files('vmbus.c'))
>>>> +system_ss.add(when: 'CONFIG_VMBUS', if_true: files('vmbus.c'))
>>>> specific_ss.add(when: 'CONFIG_SYNDBG', if_true: files('syndbg.c'))
>>>> specific_ss.add(when: 'CONFIG_HV_BALLOON', if_true: files('hv-balloon.c', 'hv-balloon-page_range_tree.c', 'hv-balloon-our_range_memslots.c'))
>>>> system_ss.add(when: 'CONFIG_HV_BALLOON', if_false: files('hv-balloon-stub.c'))
>>>
>>
>
next prev parent reply other threads:[~2025-03-07 17:28 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-03-06 6:41 [PATCH 0/7] hw/hyperv: remove duplication compilation units Pierrick Bouvier
2025-03-06 6:41 ` [PATCH 1/7] hw/hyperv/hv-balloon-stub: common compilation unit Pierrick Bouvier
2025-03-06 6:41 ` [PATCH 2/7] hw/hyperv/hyperv.h: header cleanup Pierrick Bouvier
2025-03-06 12:27 ` Alex Bennée
2025-03-06 22:40 ` Pierrick Bouvier
2025-03-06 6:41 ` [PATCH 3/7] hw/hyperv/vmbus: common compilation unit Pierrick Bouvier
2025-03-06 20:29 ` Maciej S. Szmigiero
2025-03-06 22:59 ` Pierrick Bouvier
2025-03-07 11:03 ` Maciej S. Szmigiero
2025-03-07 17:27 ` Pierrick Bouvier [this message]
2025-03-06 6:41 ` [PATCH 4/7] hw/hyperv/hyperv-proto: move SYNDBG definition from target/i386 Pierrick Bouvier
2025-03-06 10:42 ` Philippe Mathieu-Daudé
2025-03-06 6:41 ` [PATCH 5/7] hw/hyperv/syndbg: common compilation unit Pierrick Bouvier
2025-03-06 16:19 ` Richard Henderson
2025-03-06 16:23 ` Pierrick Bouvier
2025-03-06 17:58 ` Philippe Mathieu-Daudé
2025-03-06 22:56 ` Pierrick Bouvier
2025-03-07 11:07 ` Maciej S. Szmigiero
2025-03-07 14:50 ` Alex Bennée
2025-03-07 17:34 ` Maciej S. Szmigiero
2025-03-07 17:28 ` Pierrick Bouvier
2025-03-06 6:41 ` [PATCH 6/7] hw/hyperv/balloon: common balloon compilation units Pierrick Bouvier
2025-03-06 6:41 ` [PATCH 7/7] hw/hyperv/hyperv_testdev: common compilation unit Pierrick Bouvier
2025-03-06 16:26 ` [PATCH 0/7] hw/hyperv: remove duplication compilation units Richard Henderson
2025-03-06 16:35 ` Pierrick Bouvier
2025-03-06 16:35 ` Pierrick Bouvier
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=de815285-78a8-4b82-b6e3-e3315a25ed8f@linaro.org \
--to=pierrick.bouvier@linaro.org \
--cc=alex.bennee@linaro.org \
--cc=kvm@vger.kernel.org \
--cc=mail@maciej.szmigiero.name \
--cc=manos.pitsidianakis@linaro.org \
--cc=mtosatti@redhat.com \
--cc=pbonzini@redhat.com \
--cc=philmd@linaro.org \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).