[PATCH v6 13/18] accel/tcg: Do not align tb->page_addr[0]

Richard Henderson posted 18 patches 3 years, 4 months ago
Maintainers: Richard Henderson <richard.henderson@linaro.org>, Paolo Bonzini <pbonzini@redhat.com>, Eduardo Habkost <eduardo@habkost.net>, Marcel Apfelbaum <marcel.apfelbaum@gmail.com>, "Philippe Mathieu-Daudé" <f4bug@amsat.org>, Yanan Wang <wangyanan55@huawei.com>, "Alex Bennée" <alex.bennee@linaro.org>, Alexandre Iooss <erdnaxe@crans.org>, Mahmoud Mandour <ma.mandourr@gmail.com>, Peter Maydell <peter.maydell@linaro.org>, Michael Rolnik <mrolnik@gmail.com>, "Edgar E. Iglesias" <edgar.iglesias@gmail.com>, Taylor Simpson <tsimpson@quicinc.com>, Song Gao <gaosong@loongson.cn>, Xiaojuan Yang <yangxiaojuan@loongson.cn>, Laurent Vivier <laurent@vivier.eu>, Aurelien Jarno <aurelien@aurel32.net>, Jiaxun Yang <jiaxun.yang@flygoat.com>, Aleksandar Rikalo <aleksandar.rikalo@syrmia.com>, Chris Wulff <crwulff@gmail.com>, Marek Vasut <marex@denx.de>, Stafford Horne <shorne@gmail.com>, "Cédric Le Goater" <clg@kaod.org>, Daniel Henrique Barboza <danielhb413@gmail.com>, David Gibson <david@gibson.dropbear.id.au>, Greg Kurz <groug@kaod.org>, Palmer Dabbelt <palmer@dabbelt.com>, Alistair Francis <alistair.francis@wdc.com>, Bin Meng <bin.meng@windriver.com>, Yoshinori Sato <ysato@users.sourceforge.jp>, David Hildenbrand <david@redhat.com>, Cornelia Huck <cohuck@redhat.com>, Thomas Huth <thuth@redhat.com>, Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>, Artyom Tarasenko <atar4qemu@gmail.com>, Bastian Koppelmann <kbastian@mail.uni-paderborn.de>, Max Filippov <jcmvbkbc@gmail.com>, Stefan Hajnoczi <stefanha@redhat.com>
There is a newer version of this series
[PATCH v6 13/18] accel/tcg: Do not align tb->page_addr[0]
Posted by Richard Henderson 3 years, 4 months ago
Let tb->page_addr[0] contain the offset within the page of the
start of the translation block.  We need to recover this value
anyway at various points, and it is easier to discard the page
offset when it's not needed, which happens naturally via the
existing find_page shift.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/cpu-exec.c      | 16 ++++++++--------
 accel/tcg/cputlb.c        |  3 ++-
 accel/tcg/translate-all.c |  9 +++++----
 3 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 5f43b9769a..dd58a144a8 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -174,7 +174,7 @@ struct tb_desc {
     target_ulong pc;
     target_ulong cs_base;
     CPUArchState *env;
-    tb_page_addr_t phys_page1;
+    tb_page_addr_t page_addr0;
     uint32_t flags;
     uint32_t cflags;
     uint32_t trace_vcpu_dstate;
@@ -186,7 +186,7 @@ static bool tb_lookup_cmp(const void *p, const void *d)
     const struct tb_desc *desc = d;
 
     if (tb->pc == desc->pc &&
-        tb->page_addr[0] == desc->phys_page1 &&
+        tb->page_addr[0] == desc->page_addr0 &&
         tb->cs_base == desc->cs_base &&
         tb->flags == desc->flags &&
         tb->trace_vcpu_dstate == desc->trace_vcpu_dstate &&
@@ -195,8 +195,8 @@ static bool tb_lookup_cmp(const void *p, const void *d)
         if (tb->page_addr[1] == -1) {
             return true;
         } else {
-            tb_page_addr_t phys_page2;
-            target_ulong virt_page2;
+            tb_page_addr_t phys_page1;
+            target_ulong virt_page1;
 
             /*
              * We know that the first page matched, and an otherwise valid TB
@@ -207,9 +207,9 @@ static bool tb_lookup_cmp(const void *p, const void *d)
              * is different for the new TB.  Therefore any exception raised
              * here by the faulting lookup is not premature.
              */
-            virt_page2 = TARGET_PAGE_ALIGN(desc->pc);
-            phys_page2 = get_page_addr_code(desc->env, virt_page2);
-            if (tb->page_addr[1] == phys_page2) {
+            virt_page1 = TARGET_PAGE_ALIGN(desc->pc);
+            phys_page1 = get_page_addr_code(desc->env, virt_page1);
+            if (tb->page_addr[1] == phys_page1) {
                 return true;
             }
         }
@@ -235,7 +235,7 @@ static TranslationBlock *tb_htable_lookup(CPUState *cpu, target_ulong pc,
     if (phys_pc == -1) {
         return NULL;
     }
-    desc.phys_page1 = phys_pc & TARGET_PAGE_MASK;
+    desc.page_addr0 = phys_pc;
     h = tb_hash_func(phys_pc, pc, flags, cflags, *cpu->trace_dstate);
     return qht_lookup_custom(&tb_ctx.htable, &desc, h, tb_lookup_cmp);
 }
diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index 361078471b..a0db2d32a8 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -951,7 +951,8 @@ void tlb_flush_page_bits_by_mmuidx_all_cpus_synced(CPUState *src_cpu,
    can be detected */
 void tlb_protect_code(ram_addr_t ram_addr)
 {
-    cpu_physical_memory_test_and_clear_dirty(ram_addr, TARGET_PAGE_SIZE,
+    cpu_physical_memory_test_and_clear_dirty(ram_addr & TARGET_PAGE_MASK,
+                                             TARGET_PAGE_SIZE,
                                              DIRTY_MEMORY_CODE);
 }
 
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index ca685f6ede..3a63113c41 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -1167,7 +1167,7 @@ static void do_tb_phys_invalidate(TranslationBlock *tb, bool rm_from_page_list)
     qemu_spin_unlock(&tb->jmp_lock);
 
     /* remove the TB from the hash list */
-    phys_pc = tb->page_addr[0] + (tb->pc & ~TARGET_PAGE_MASK);
+    phys_pc = tb->page_addr[0];
     h = tb_hash_func(phys_pc, tb->pc, tb->flags, orig_cflags,
                      tb->trace_vcpu_dstate);
     if (!qht_remove(&tb_ctx.htable, tb, h)) {
@@ -1291,7 +1291,7 @@ tb_link_page(TranslationBlock *tb, tb_page_addr_t phys_pc,
      * we can only insert TBs that are fully initialized.
      */
     page_lock_pair(&p, phys_pc, &p2, phys_page2, true);
-    tb_page_add(p, tb, 0, phys_pc & TARGET_PAGE_MASK);
+    tb_page_add(p, tb, 0, phys_pc);
     if (p2) {
         tb_page_add(p2, tb, 1, phys_page2);
     } else {
@@ -1644,11 +1644,12 @@ tb_invalidate_phys_page_range__locked(struct page_collection *pages,
         if (n == 0) {
             /* NOTE: tb_end may be after the end of the page, but
                it is not a problem */
-            tb_start = tb->page_addr[0] + (tb->pc & ~TARGET_PAGE_MASK);
+            tb_start = tb->page_addr[0];
             tb_end = tb_start + tb->size;
         } else {
             tb_start = tb->page_addr[1];
-            tb_end = tb_start + ((tb->pc + tb->size) & ~TARGET_PAGE_MASK);
+            tb_end = tb_start + ((tb->page_addr[0] + tb->size)
+                                 & ~TARGET_PAGE_MASK);
         }
         if (!(tb_end <= start || tb_start >= end)) {
 #ifdef TARGET_HAS_PRECISE_SMC
-- 
2.34.1
Re: [PATCH v6 13/18] accel/tcg: Do not align tb->page_addr[0]
Posted by Alex Bennée 3 years, 4 months ago
Richard Henderson <richard.henderson@linaro.org> writes:

> Let tb->page_addr[0] contain the offset within the page of the
> start of the translation block.  We need to recover this value
> anyway at various points, and it is easier to discard the page
> offset when it's not needed, which happens naturally via the
> existing find_page shift.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  accel/tcg/cpu-exec.c      | 16 ++++++++--------
>  accel/tcg/cputlb.c        |  3 ++-
>  accel/tcg/translate-all.c |  9 +++++----
>  3 files changed, 15 insertions(+), 13 deletions(-)
>
> diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
> index 5f43b9769a..dd58a144a8 100644
> --- a/accel/tcg/cpu-exec.c
> +++ b/accel/tcg/cpu-exec.c
> @@ -174,7 +174,7 @@ struct tb_desc {
>      target_ulong pc;
>      target_ulong cs_base;
>      CPUArchState *env;
> -    tb_page_addr_t phys_page1;
> +    tb_page_addr_t page_addr0;

We don't actually document that this is an offset here (or indeed in
TranslationBlock) and the definition of tb_page_addr_t:

  /* Page tracking code uses ram addresses in system mode, and virtual
     addresses in userspace mode.  Define tb_page_addr_t to be an appropriate
     type.  */
  #if defined(CONFIG_USER_ONLY)
  typedef abi_ulong tb_page_addr_t;
  #define TB_PAGE_ADDR_FMT TARGET_ABI_FMT_lx
  #else
  typedef ram_addr_t tb_page_addr_t;
  #define TB_PAGE_ADDR_FMT RAM_ADDR_FMT
  #endif

implies these are full size pointers into the guests address space.
Either we need a new type (tb_page_offset_t) or to properly comment the
structures with what they mean.

Otherwise:

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée
Re: [PATCH v6 13/18] accel/tcg: Do not align tb->page_addr[0]
Posted by Richard Henderson 3 years, 4 months ago
On 10/3/22 05:47, Alex Bennée wrote:
> 
> Richard Henderson <richard.henderson@linaro.org> writes:
> 
>> Let tb->page_addr[0] contain the offset within the page of the
>> start of the translation block.  We need to recover this value
>> anyway at various points, and it is easier to discard the page
>> offset when it's not needed, which happens naturally via the
>> existing find_page shift.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> ---
>>   accel/tcg/cpu-exec.c      | 16 ++++++++--------
>>   accel/tcg/cputlb.c        |  3 ++-
>>   accel/tcg/translate-all.c |  9 +++++----
>>   3 files changed, 15 insertions(+), 13 deletions(-)
>>
>> diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
>> index 5f43b9769a..dd58a144a8 100644
>> --- a/accel/tcg/cpu-exec.c
>> +++ b/accel/tcg/cpu-exec.c
>> @@ -174,7 +174,7 @@ struct tb_desc {
>>       target_ulong pc;
>>       target_ulong cs_base;
>>       CPUArchState *env;
>> -    tb_page_addr_t phys_page1;
>> +    tb_page_addr_t page_addr0;
> 
> We don't actually document that this is an offset here (or indeed in
> TranslationBlock) and the definition of tb_page_addr_t:
> 
>    /* Page tracking code uses ram addresses in system mode, and virtual
>       addresses in userspace mode.  Define tb_page_addr_t to be an appropriate
>       type.  */
>    #if defined(CONFIG_USER_ONLY)
>    typedef abi_ulong tb_page_addr_t;
>    #define TB_PAGE_ADDR_FMT TARGET_ABI_FMT_lx
>    #else
>    typedef ram_addr_t tb_page_addr_t;
>    #define TB_PAGE_ADDR_FMT RAM_ADDR_FMT
>    #endif
> 
> implies these are full size pointers into the guests address space.

And that's what I've got.  What we we were storing in phys_page1 before was a full size 
pointer that was page aligned.  I'm now dropping the page alignment and having a full size 
pointer to the exact first byte of the translated code.

Is that clearer?  How would you improve the wording?


r~

> Either we need a new type (tb_page_offset_t) or to properly comment the
> structures with what they mean.
> 
> Otherwise:
> 
> Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
> 


Re: [PATCH v6 13/18] accel/tcg: Do not align tb->page_addr[0]
Posted by Alex Bennée 3 years, 4 months ago
Richard Henderson <richard.henderson@linaro.org> writes:

> On 10/3/22 05:47, Alex Bennée wrote:
>> Richard Henderson <richard.henderson@linaro.org> writes:
>> 
>>> Let tb->page_addr[0] contain the offset within the page of the
>>> start of the translation block.  We need to recover this value
>>> anyway at various points, and it is easier to discard the page
>>> offset when it's not needed, which happens naturally via the
>>> existing find_page shift.
>>>
>>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>>> ---
>>>   accel/tcg/cpu-exec.c      | 16 ++++++++--------
>>>   accel/tcg/cputlb.c        |  3 ++-
>>>   accel/tcg/translate-all.c |  9 +++++----
>>>   3 files changed, 15 insertions(+), 13 deletions(-)
>>>
>>> diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
>>> index 5f43b9769a..dd58a144a8 100644
>>> --- a/accel/tcg/cpu-exec.c
>>> +++ b/accel/tcg/cpu-exec.c
>>> @@ -174,7 +174,7 @@ struct tb_desc {
>>>       target_ulong pc;
>>>       target_ulong cs_base;
>>>       CPUArchState *env;
>>> -    tb_page_addr_t phys_page1;
>>> +    tb_page_addr_t page_addr0;
>> We don't actually document that this is an offset here (or indeed in
>> TranslationBlock) and the definition of tb_page_addr_t:
>>    /* Page tracking code uses ram addresses in system mode, and
>> virtual
>>       addresses in userspace mode.  Define tb_page_addr_t to be an appropriate
>>       type.  */
>>    #if defined(CONFIG_USER_ONLY)
>>    typedef abi_ulong tb_page_addr_t;
>>    #define TB_PAGE_ADDR_FMT TARGET_ABI_FMT_lx
>>    #else
>>    typedef ram_addr_t tb_page_addr_t;
>>    #define TB_PAGE_ADDR_FMT RAM_ADDR_FMT
>>    #endif
>> implies these are full size pointers into the guests address space.
>
> And that's what I've got.  What we we were storing in phys_page1
> before was a full size pointer that was page aligned.  I'm now
> dropping the page alignment and having a full size pointer to the
> exact first byte of the translated code.

OK then I'm confused by the commit message which says:

  Let tb->page_addr[0] contain the offset within the page of the
  start of the translation block

> Is that clearer?  How would you improve the wording?
>
>
> r~
>
>> Either we need a new type (tb_page_offset_t) or to properly comment the
>> structures with what they mean.
>> Otherwise:
>> Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
>> 


-- 
Alex Bennée