From: David Hildenbrand <david@redhat.com>
To: Thomas Huth <thuth@redhat.com>, qemu-devel@nongnu.org
Cc: Claudio Imbrenda <imbrenda@linux.ibm.com>,
qemu-s390x@nongnu.org, Cornelia Huck <cohuck@redhat.com>,
Richard Henderson <richard.henderson@linaro.org>,
Janosch Frank <frankja@linux.ibm.com>
Subject: Re: [PATCH v4] target/s390x: Implement the MVPG condition-code-option bit
Date: Wed, 3 Mar 2021 20:39:25 +0100 [thread overview]
Message-ID: <c938bfb6-7322-1738-8492-972b83cb7c99@redhat.com> (raw)
In-Reply-To: <20210303132850.459687-1-thuth@redhat.com>
On 03.03.21 14:28, Thomas Huth wrote:
> From: Richard Henderson <richard.henderson@linaro.org>
>
> 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.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> [thuth: Added logic to still inject protection exceptions]
> Signed-off-by: Thomas Huth <thuth@redhat.com>
> ---
> v4: Add logic to inject protection exceptions if necessary
>
> target/s390x/cpu.h | 3 ++
> target/s390x/excp_helper.c | 3 ++
> target/s390x/mem_helper.c | 93 ++++++++++++++++++++++++++++----------
> 3 files changed, 76 insertions(+), 23 deletions(-)
>
> diff --git a/target/s390x/cpu.h b/target/s390x/cpu.h
> index 60d434d5ed..825503c6c0 100644
> --- a/target/s390x/cpu.h
> +++ b/target/s390x/cpu.h
> @@ -114,6 +114,9 @@ struct CPUS390XState {
>
> uint64_t diag318_info;
>
Should we start wrapping that stuff into #ifdef CONFIG_TCG ?
> + uint64_t tlb_fill_tec; /* translation exception code during tlb_fill */
> + int tlb_fill_exc; /* exception number seen during tlb_fill */
> +
> /* Fields up to this point are cleared by a CPU reset */
> struct {} end_reset_fields;
>
> diff --git a/target/s390x/excp_helper.c b/target/s390x/excp_helper.c
> index ce16af394b..c48cd6b46f 100644
> --- a/target/s390x/excp_helper.c
> +++ b/target/s390x/excp_helper.c
> @@ -164,6 +164,9 @@ bool s390_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
> tec = 0; /* unused */
> }
>
> + env->tlb_fill_exc = excp;
> + env->tlb_fill_tec = tec;
> +
Just what I had in mind.
> if (!excp) {
> qemu_log_mask(CPU_LOG_MMU,
> "%s: set tlb %" PRIx64 " -> %" PRIx64 " (%x)\n",
> diff --git a/target/s390x/mem_helper.c b/target/s390x/mem_helper.c
> index 25cfede806..cf741541d3 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;
^ I recall PAGE_WRITE_INV handling where we immediately set
TLB_INVALID_MASK again on write access (to handle low-address protection
cleanly). I suspect that TLB_INVALID_MASK will be set in that case (I
could be wrong, though).
What certainly would work is checking for "haddr != NULL".
/* Don't rely on TLB_INVALID_MASK - see PAGE_WRITE_INV handling. */
if (unlikely(!haddr1)) {
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);
> + }
> +
[...]
> /* 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 +894,24 @@ 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) {
> + if (env->tlb_fill_exc == PGM_PROTECTION) {
> + stq_phys(env_cpu(env)->as,
> + env->psa + offsetof(LowCore, trans_exc_code),
> + env->tlb_fill_tec);
> + tcg_s390_program_interrupt(env, PGM_PROTECTION, ra);
> + }
> + return 1;
> + }
> access_memmove(env, &desta, &srca, ra);
> return 0; /* data moved */
> }
>
Apart from that, looks good to me.
--
Thanks,
David / dhildenb
next prev parent reply other threads:[~2021-03-03 19:40 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-03 13:28 [PATCH v4] target/s390x: Implement the MVPG condition-code-option bit Thomas Huth
2021-03-03 18:22 ` Richard Henderson
2021-03-03 19:39 ` David Hildenbrand [this message]
2021-03-03 21:05 ` Richard Henderson
2021-03-03 21:11 ` David Hildenbrand
2021-03-03 21:19 ` Richard Henderson
2021-03-03 21:22 ` David Hildenbrand
2021-03-03 21:36 ` Richard Henderson
2021-03-04 8:10 ` David Hildenbrand
2021-03-04 8:17 ` Cornelia Huck
2021-03-09 21:05 ` Thomas Huth
2021-03-10 20:49 ` David Hildenbrand
2021-03-11 10:21 ` Cornelia Huck
2021-03-11 14:03 ` David Hildenbrand
2021-03-11 15:58 ` Richard Henderson
2021-03-11 16:00 ` David Hildenbrand
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=c938bfb6-7322-1738-8492-972b83cb7c99@redhat.com \
--to=david@redhat.com \
--cc=cohuck@redhat.com \
--cc=frankja@linux.ibm.com \
--cc=imbrenda@linux.ibm.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-s390x@nongnu.org \
--cc=richard.henderson@linaro.org \
--cc=thuth@redhat.com \
/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).