From patchwork Wed Dec 6 22:40:53 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Song Liu X-Patchwork-Id: 13482349 X-Patchwork-Delegate: bpf@iogearbox.net Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 91C492E40D for ; Wed, 6 Dec 2023 22:42:03 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="ZdYxN/kX" Received: by smtp.kernel.org (Postfix) with ESMTPSA id A4A57C433C7; Wed, 6 Dec 2023 22:42:01 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1701902523; bh=MH2YRkumw1iokydCI0zgZbVeI2huQ/78l+SGMVOnKRk=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=ZdYxN/kXLubqcRkfGkNznRtRaICa1JPt0DY9997IHDEDFva84r+yhWP3wYbfx8qWt 7B4ITab45IV4E3CGmaBYWxa1SbZbwwRpRaKlAEEWdiP53khdysT1e/cGfZyfJT9zo0 3hQ+WieYWVt6qwPS4aruh0OOWVRDBIcPUN/RxdPgIS3ApQ5SdMB+sbNUOn0lcVe5RL 5PUep2AqV5+e2jAY0p6MjoV76BXZwZTL6rc2ZKcLHb3NhYgj7sutcHhXlYUHRtLipp loeChpYhr5FqN2PQ8sbisRbCR2ISJOZBU5yKToB0StH46MXj9GNzKv6OdIBasaKq0W 6Tq8UAI5xVuEw== From: Song Liu To: bpf@vger.kernel.org Cc: ast@kernel.org, daniel@iogearbox.net, andrii@kernel.org, martin.lau@linux.dev, kernel-team@meta.com, Song Liu , Ilya Leoshkevich , Jiri Olsa , =?utf-8?b?QmrDtnJuIFTDtnBlbA==?= Subject: [PATCH v7 bpf-next 6/7] bpf: Use arch_bpf_trampoline_size Date: Wed, 6 Dec 2023 14:40:53 -0800 Message-Id: <20231206224054.492250-7-song@kernel.org> X-Mailer: git-send-email 2.34.1 In-Reply-To: <20231206224054.492250-1-song@kernel.org> References: <20231206224054.492250-1-song@kernel.org> Precedence: bulk X-Mailing-List: bpf@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: bpf@iogearbox.net Instead of blindly allocating PAGE_SIZE for each trampoline, check the size of the trampoline with arch_bpf_trampoline_size(). This size is saved in bpf_tramp_image->size, and used for modmem charge/uncharge. The fallback arch_alloc_bpf_trampoline() still allocates a whole page because we need to use set_memory_* to protect the memory. struct_ops trampoline still uses a whole page for multiple trampolines. With this size check at caller (regular trampoline and struct_ops trampoline), remove arch_bpf_trampoline_size() from arch_prepare_bpf_trampoline() in archs. Also, update bpf_image_ksym_add() to handle symbol of different sizes. Signed-off-by: Song Liu Acked-by: Ilya Leoshkevich Tested-by: Ilya Leoshkevich # on s390x Acked-by: Jiri Olsa Acked-by: Björn Töpel Tested-by: Björn Töpel # on riscv --- arch/arm64/net/bpf_jit_comp.c | 7 ----- arch/riscv/net/bpf_jit_comp64.c | 7 ----- include/linux/bpf.h | 3 +- kernel/bpf/bpf_struct_ops.c | 7 +++++ kernel/bpf/dispatcher.c | 2 +- kernel/bpf/trampoline.c | 55 ++++++++++++++++++++------------- 6 files changed, 44 insertions(+), 37 deletions(-) diff --git a/arch/arm64/net/bpf_jit_comp.c b/arch/arm64/net/bpf_jit_comp.c index a6671253b7ed..8955da5c47cf 100644 --- a/arch/arm64/net/bpf_jit_comp.c +++ b/arch/arm64/net/bpf_jit_comp.c @@ -2079,13 +2079,6 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, if (nregs > 8) return -ENOTSUPP; - ret = arch_bpf_trampoline_size(m, flags, tlinks, func_addr); - if (ret < 0) - return ret; - - if (ret > ((long)image_end - (long)image)) - return -EFBIG; - jit_fill_hole(image, (unsigned int)(image_end - image)); ret = prepare_trampoline(&ctx, im, tlinks, func_addr, nregs, flags); diff --git a/arch/riscv/net/bpf_jit_comp64.c b/arch/riscv/net/bpf_jit_comp64.c index 35747fafde57..58dc64dd94a8 100644 --- a/arch/riscv/net/bpf_jit_comp64.c +++ b/arch/riscv/net/bpf_jit_comp64.c @@ -1052,13 +1052,6 @@ int arch_prepare_bpf_trampoline(struct bpf_tramp_image *im, void *image, int ret; struct rv_jit_context ctx; - ret = arch_bpf_trampoline_size(im, m, flags, tlinks, func_addr); - if (ret < 0) - return ret; - - if (ret > (long)image_end - (long)image) - return -EFBIG; - ctx.ninsns = 0; /* * The bpf_int_jit_compile() uses a RW buffer (ctx.insns) to write the diff --git a/include/linux/bpf.h b/include/linux/bpf.h index 2332ddeb396b..c1a06263a4f3 100644 --- a/include/linux/bpf.h +++ b/include/linux/bpf.h @@ -1141,6 +1141,7 @@ enum bpf_tramp_prog_type { struct bpf_tramp_image { void *image; + int size; struct bpf_ksym ksym; struct percpu_ref pcref; void *ip_after_call; @@ -1325,7 +1326,7 @@ int arch_prepare_bpf_dispatcher(void *image, void *buf, s64 *funcs, int num_func void bpf_dispatcher_change_prog(struct bpf_dispatcher *d, struct bpf_prog *from, struct bpf_prog *to); /* Called only from JIT-enabled code, so there's no need for stubs. */ -void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym); +void bpf_image_ksym_add(void *data, unsigned int size, struct bpf_ksym *ksym); void bpf_image_ksym_del(struct bpf_ksym *ksym); void bpf_ksym_add(struct bpf_ksym *ksym); void bpf_ksym_del(struct bpf_ksym *ksym); diff --git a/kernel/bpf/bpf_struct_ops.c b/kernel/bpf/bpf_struct_ops.c index e9e95879bce2..4d53c53fc5aa 100644 --- a/kernel/bpf/bpf_struct_ops.c +++ b/kernel/bpf/bpf_struct_ops.c @@ -355,6 +355,7 @@ int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks, void *image, void *image_end) { u32 flags; + int size; tlinks[BPF_TRAMP_FENTRY].links[0] = link; tlinks[BPF_TRAMP_FENTRY].nr_links = 1; @@ -362,6 +363,12 @@ int bpf_struct_ops_prepare_trampoline(struct bpf_tramp_links *tlinks, * and it must be used alone. */ flags = model->ret_size > 0 ? BPF_TRAMP_F_RET_FENTRY_RET : 0; + + size = arch_bpf_trampoline_size(model, flags, tlinks, NULL); + if (size < 0) + return size; + if (size > (unsigned long)image_end - (unsigned long)image) + return -E2BIG; return arch_prepare_bpf_trampoline(NULL, image, image_end, model, flags, tlinks, NULL); } diff --git a/kernel/bpf/dispatcher.c b/kernel/bpf/dispatcher.c index 56760fc10e78..70fb82bf1637 100644 --- a/kernel/bpf/dispatcher.c +++ b/kernel/bpf/dispatcher.c @@ -154,7 +154,7 @@ void bpf_dispatcher_change_prog(struct bpf_dispatcher *d, struct bpf_prog *from, d->image = NULL; goto out; } - bpf_image_ksym_add(d->image, &d->ksym); + bpf_image_ksym_add(d->image, PAGE_SIZE, &d->ksym); } prev_num_progs = d->num_progs; diff --git a/kernel/bpf/trampoline.c b/kernel/bpf/trampoline.c index b553cbd89e55..d382f5ebe06c 100644 --- a/kernel/bpf/trampoline.c +++ b/kernel/bpf/trampoline.c @@ -115,10 +115,10 @@ bool bpf_prog_has_trampoline(const struct bpf_prog *prog) (ptype == BPF_PROG_TYPE_LSM && eatype == BPF_LSM_MAC); } -void bpf_image_ksym_add(void *data, struct bpf_ksym *ksym) +void bpf_image_ksym_add(void *data, unsigned int size, struct bpf_ksym *ksym) { ksym->start = (unsigned long) data; - ksym->end = ksym->start + PAGE_SIZE; + ksym->end = ksym->start + size; bpf_ksym_add(ksym); perf_event_ksymbol(PERF_RECORD_KSYMBOL_TYPE_BPF, ksym->start, PAGE_SIZE, false, ksym->name); @@ -254,8 +254,8 @@ bpf_trampoline_get_progs(const struct bpf_trampoline *tr, int *total, bool *ip_a static void bpf_tramp_image_free(struct bpf_tramp_image *im) { bpf_image_ksym_del(&im->ksym); - arch_free_bpf_trampoline(im->image, PAGE_SIZE); - bpf_jit_uncharge_modmem(PAGE_SIZE); + arch_free_bpf_trampoline(im->image, im->size); + bpf_jit_uncharge_modmem(im->size); percpu_ref_exit(&im->pcref); kfree_rcu(im, rcu); } @@ -349,7 +349,7 @@ static void bpf_tramp_image_put(struct bpf_tramp_image *im) call_rcu_tasks_trace(&im->rcu, __bpf_tramp_image_put_rcu_tasks); } -static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key) +static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key, int size) { struct bpf_tramp_image *im; struct bpf_ksym *ksym; @@ -360,12 +360,13 @@ static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key) if (!im) goto out; - err = bpf_jit_charge_modmem(PAGE_SIZE); + err = bpf_jit_charge_modmem(size); if (err) goto out_free_im; + im->size = size; err = -ENOMEM; - im->image = image = arch_alloc_bpf_trampoline(PAGE_SIZE); + im->image = image = arch_alloc_bpf_trampoline(size); if (!image) goto out_uncharge; @@ -376,13 +377,13 @@ static struct bpf_tramp_image *bpf_tramp_image_alloc(u64 key) ksym = &im->ksym; INIT_LIST_HEAD_RCU(&ksym->lnode); snprintf(ksym->name, KSYM_NAME_LEN, "bpf_trampoline_%llu", key); - bpf_image_ksym_add(image, ksym); + bpf_image_ksym_add(image, size, ksym); return im; out_free_image: - arch_free_bpf_trampoline(im->image, PAGE_SIZE); + arch_free_bpf_trampoline(im->image, im->size); out_uncharge: - bpf_jit_uncharge_modmem(PAGE_SIZE); + bpf_jit_uncharge_modmem(size); out_free_im: kfree(im); out: @@ -395,7 +396,7 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut struct bpf_tramp_links *tlinks; u32 orig_flags = tr->flags; bool ip_arg = false; - int err, total; + int err, total, size; tlinks = bpf_trampoline_get_progs(tr, &total, &ip_arg); if (IS_ERR(tlinks)) @@ -408,12 +409,6 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut goto out; } - im = bpf_tramp_image_alloc(tr->key); - if (IS_ERR(im)) { - err = PTR_ERR(im); - goto out; - } - /* clear all bits except SHARE_IPMODIFY and TAIL_CALL_CTX */ tr->flags &= (BPF_TRAMP_F_SHARE_IPMODIFY | BPF_TRAMP_F_TAIL_CALL_CTX); @@ -437,13 +432,31 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut tr->flags |= BPF_TRAMP_F_ORIG_STACK; #endif - err = arch_prepare_bpf_trampoline(im, im->image, im->image + PAGE_SIZE, + size = arch_bpf_trampoline_size(&tr->func.model, tr->flags, + tlinks, tr->func.addr); + if (size < 0) { + err = size; + goto out; + } + + if (size > PAGE_SIZE) { + err = -E2BIG; + goto out; + } + + im = bpf_tramp_image_alloc(tr->key, size); + if (IS_ERR(im)) { + err = PTR_ERR(im); + goto out; + } + + err = arch_prepare_bpf_trampoline(im, im->image, im->image + size, &tr->func.model, tr->flags, tlinks, tr->func.addr); if (err < 0) goto out_free; - arch_protect_bpf_trampoline(im->image, PAGE_SIZE); + arch_protect_bpf_trampoline(im->image, im->size); WARN_ON(tr->cur_image && total == 0); if (tr->cur_image) @@ -463,8 +476,8 @@ static int bpf_trampoline_update(struct bpf_trampoline *tr, bool lock_direct_mut tr->fops->func = NULL; tr->fops->trampoline = 0; - /* reset im->image memory attr for arch_prepare_bpf_trampoline */ - arch_unprotect_bpf_trampoline(im->image, PAGE_SIZE); + /* free im memory and reallocate later */ + bpf_tramp_image_free(im); goto again; } #endif