qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Thomas Huth <thuth@redhat.com>
To: Richard Henderson <richard.henderson@linaro.org>, qemu-devel@nongnu.org
Cc: qemu-s390x@nongnu.org, David Hildenbrand <david@redhat.com>
Subject: Re: [PATCH v3] target/s390x: Implement the MVPG condition-code-option bit
Date: Wed, 3 Mar 2021 12:36:06 +0100	[thread overview]
Message-ID: <fb6b66d2-1764-174f-138b-d26b4a3bb64f@redhat.com> (raw)
In-Reply-To: <82ea6a14-af12-818c-622b-4ecf06905967@redhat.com>

On 03/03/2021 07.25, Thomas Huth wrote:
> On 03/03/2021 03.44, Richard Henderson wrote:
>> If the CCO bit is set, MVPG should not generate an exception but
>> report page translation faults via a CC code.
>>
>> Create a new helper, access_prepare_nf, which can use probe_access_flags
>> in non-faulting mode, and then handle watchpoints.
>>
>> Cc: David Hildenbrand <david@redhat.com>
>> Reported-by: Thomas Huth <thuth@redhat.com>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
>> ---
>>   target/s390x/mem_helper.c | 87 ++++++++++++++++++++++++++++-----------
>>   1 file changed, 64 insertions(+), 23 deletions(-)
>>
>> diff --git a/target/s390x/mem_helper.c b/target/s390x/mem_helper.c
>> index 25cfede806..b397333c0b 100644
>> --- a/target/s390x/mem_helper.c
>> +++ b/target/s390x/mem_helper.c
>> @@ -130,28 +130,62 @@ typedef struct S390Access {
>>       int mmu_idx;
>>   } S390Access;
>> +static bool access_prepare_nf(S390Access *access, CPUS390XState *env,
>> +                              bool nofault, vaddr vaddr1, int size,
>> +                              MMUAccessType access_type,
>> +                              int mmu_idx, uintptr_t ra)
>> +{
>> +    void *haddr1, *haddr2 = NULL;
>> +    int size1, size2;
>> +    vaddr vaddr2 = 0;
>> +    int flags;
>> +
>> +    assert(size > 0 && size <= 4096);
>> +
>> +    size1 = MIN(size, -(vaddr1 | TARGET_PAGE_MASK)),
>> +    size2 = size - size1;
>> +
>> +    flags = probe_access_flags(env, vaddr1, access_type, mmu_idx,
>> +                               nofault, &haddr1, ra);
>> +    if (unlikely(size2)) {
>> +        /* The access crosses page boundaries. */
>> +        vaddr2 = wrap_address(env, vaddr1 + size1);
>> +        flags |= probe_access_flags(env, vaddr2, access_type, mmu_idx,
>> +                                    nofault, &haddr2, ra);
>> +    }
>> +
>> +    if (unlikely(flags & TLB_INVALID_MASK)) {
>> +        return false;
>> +    }
>> +    if (unlikely(flags & TLB_WATCHPOINT)) {
>> +        /* S390 does not presently use transaction attributes. */
>> +        cpu_check_watchpoint(env_cpu(env), vaddr1, size,
>> +                             MEMTXATTRS_UNSPECIFIED,
>> +                             (access_type == MMU_DATA_STORE
>> +                              ? BP_MEM_WRITE : BP_MEM_READ), ra);
>> +    }
>> +
>> +    *access = (S390Access) {
>> +        .vaddr1 = vaddr1,
>> +        .vaddr2 = vaddr2,
>> +        .haddr1 = haddr1,
>> +        .haddr2 = haddr2,
>> +        .size1 = size1,
>> +        .size2 = size2,
>> +        .mmu_idx = mmu_idx
>> +    };
>> +    return true;
>> +}
>> +
>>   static S390Access access_prepare(CPUS390XState *env, vaddr vaddr, int size,
>>                                    MMUAccessType access_type, int mmu_idx,
>>                                    uintptr_t ra)
>>   {
>> -    S390Access access = {
>> -        .vaddr1 = vaddr,
>> -        .size1 = MIN(size, -(vaddr | TARGET_PAGE_MASK)),
>> -        .mmu_idx = mmu_idx,
>> -    };
>> -
>> -    g_assert(size > 0 && size <= 4096);
>> -    access.haddr1 = probe_access(env, access.vaddr1, access.size1, 
>> access_type,
>> -                                 mmu_idx, ra);
>> -
>> -    if (unlikely(access.size1 != size)) {
>> -        /* The access crosses page boundaries. */
>> -        access.vaddr2 = wrap_address(env, vaddr + access.size1);
>> -        access.size2 = size - access.size1;
>> -        access.haddr2 = probe_access(env, access.vaddr2, access.size2,
>> -                                     access_type, mmu_idx, ra);
>> -    }
>> -    return access;
>> +    S390Access ret;
>> +    bool ok = access_prepare_nf(&ret, env, false, vaddr, size,
>> +                                access_type, mmu_idx, ra);
>> +    assert(ok);
>> +    return ret;
>>   }
>>   /* Helper to handle memset on a single page. */
>> @@ -845,8 +879,10 @@ uint32_t HELPER(mvpg)(CPUS390XState *env, uint64_t 
>> r0, uint64_t r1, uint64_t r2)
>>       const int mmu_idx = cpu_mmu_index(env, false);
>>       const bool f = extract64(r0, 11, 1);
>>       const bool s = extract64(r0, 10, 1);
>> +    const bool cco = extract64(r0, 8, 1);
>>       uintptr_t ra = GETPC();
>>       S390Access srca, desta;
>> +    bool ok;
>>       if ((f && s) || extract64(r0, 12, 4)) {
>>           tcg_s390_program_interrupt(env, PGM_SPECIFICATION, GETPC());
>> @@ -858,13 +894,18 @@ uint32_t HELPER(mvpg)(CPUS390XState *env, uint64_t 
>> r0, uint64_t r1, uint64_t r2)
>>       /*
>>        * TODO:
>>        * - Access key handling
>> -     * - CC-option with surpression of page-translation exceptions
>>        * - Store r1/r2 register identifiers at real location 162
>>        */
>> -    srca = access_prepare(env, r2, TARGET_PAGE_SIZE, MMU_DATA_LOAD, mmu_idx,
>> -                          ra);
>> -    desta = access_prepare(env, r1, TARGET_PAGE_SIZE, MMU_DATA_STORE, 
>> mmu_idx,
>> -                           ra);
>> +    ok = access_prepare_nf(&srca, env, cco, r2, TARGET_PAGE_SIZE,
>> +                           MMU_DATA_LOAD, mmu_idx, ra);
>> +    if (!ok) {
>> +        return 2;
>> +    }
>> +    ok = access_prepare_nf(&desta, env, cco, r1, TARGET_PAGE_SIZE,
>> +                           MMU_DATA_STORE, mmu_idx, ra);
>> +    if (!ok) {
>> +        return 1;
>> +    }
> 
> Thanks, this looks promising, but one of the MVPG kvm-unit-tests is still 
> failing with this patch - the one that checks for an exception if the 
> destination page is marked as read-only. MVPG only returns CC1 for invalid 
> page table entries - but if the page is write-protected, it still causes a 
> protection exception. That's why I've been checking "if (exc && exc != 
> PGM_PROTECTION)" in my version of the patch.

FWIW, I can get the MVPG kvm-unit-test working with your patch if I add
this on top:

diff a/target/s390x/mem_helper.c b/target/s390x/mem_helper.c
--- a/target/s390x/mem_helper.c
+++ b/target/s390x/mem_helper.c
@@ -904,7 +904,14 @@ uint32_t HELPER(mvpg)(CPUS390XState *env, uint64_t r0, uint64_t r1, uint64_t r2)
      ok = access_prepare_nf(&desta, env, cco, r1, TARGET_PAGE_SIZE,
                             MMU_DATA_STORE, mmu_idx, ra);
      if (!ok) {
-        return 1;
+        ok = access_prepare_nf(&desta, env, cco, r1, TARGET_PAGE_SIZE,
+                               MMU_DATA_LOAD, mmu_idx, ra);
+        if (!ok) {
+            return 1;
+        }
+        /* If reading was ok, then the page must be protected. */
+        /* TODO: Set a translation exception code in lowcore? */
+        tcg_s390_program_interrupt(env, PGM_PROTECTION, ra);
      }
      access_memmove(env, &desta, &srca, ra);
      return 0; /* data moved */

... yeah, it's ugly to call access_prepare_nf() again with MMU_DATA_LOAD,
and it's still missing the translation exception code ... but at least
the kvm-unit-test is happy that way...

  Thomas
   



  reply	other threads:[~2021-03-03 11:37 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-03  2:44 [PATCH v3] target/s390x: Implement the MVPG condition-code-option bit Richard Henderson
2021-03-03  6:25 ` Thomas Huth
2021-03-03 11:36   ` Thomas Huth [this message]
2021-03-03 11:40     ` David Hildenbrand
2021-03-03 12:32       ` Thomas Huth

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=fb6b66d2-1764-174f-138b-d26b4a3bb64f@redhat.com \
    --to=thuth@redhat.com \
    --cc=david@redhat.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-s390x@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).