From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from foss.arm.com (foss.arm.com [217.140.110.172]) by smtp.subspace.kernel.org (Postfix) with ESMTP id 40E8B3BFE3E; Tue, 29 Sep 2026 22:25:08 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=217.140.110.172 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1790720710; cv=none; b=FJBHpBRd+novhY4yW8Cbd2xURYZJDaEmY8a6KUHzD5OLAl4DrxwPNb/ISNazFW35kDa02C0KV13HbrH/tzReLmdmPMU7pTCEuR3x+lYngY5i6JssjBmdech7rvib+9vLdflghTbUq1C5M9cPVIUqdIb1FoR+itS/jBoeItxpmVk= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1790720710; c=relaxed/simple; bh=PSFY58zfgo7NCqRkFalHFfkLib+HJnqBr347USh2gQk=; h=Message-ID:Date:MIME-Version:Subject:To:Cc:References:From: In-Reply-To:Content-Type; b=epGCc3Clb7HoiaoBqwpz/itup8UTSltlqhPzJih7G+PegZ9wPpu7FUbsdNuHuF5C1aV1HaVITlEFIPPbq46vnst5KwPf8bAUtX9WTBnG5b2lQve0Wu/l/lfn9A8jdb4Hw1Is/cZKPGWQUloFCQRcI2pyQ2OtRJASe5jsibNDEZA= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=arm.com; spf=pass smtp.mailfrom=arm.com; dkim=pass (1024-bit key) header.d=arm.com header.i=@arm.com header.b=m7hQcXa/; arc=none smtp.client-ip=217.140.110.172 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=arm.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=arm.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=arm.com header.i=@arm.com header.b="m7hQcXa/" Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 3528C1477; Tue, 29 Sep 2026 15:25:04 -0700 (PDT) Received: from [10.57.11.173] (unknown [10.57.11.173]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id 80E343F85F; Tue, 29 Sep 2026 15:25:04 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=simple/simple; d=arm.com; s=foss; t=1790720707; bh=PSFY58zfgo7NCqRkFalHFfkLib+HJnqBr347USh2gQk=; h=Date:Subject:To:Cc:References:From:In-Reply-To:From; b=m7hQcXa/IwCt+CQ2zrTCDaxuzTtR7RybPJ6LZhfLPRxIihumkuxswsX8/5Fw0/5b6 4WxOaQ1WT4kuoYRhRgrBFiBkHeAKPTuQY5zbjMANIaBu+ua2+pEWw18fZIw8iUzy2Z uq6US/z6I6kHX1t+zYCc9gmUqNkA/dzMI7SMDSiw= Message-ID: <7859d7ee-e9cd-4589-9818-7eaafe28785b@arm.com> Date: Tue, 29 Sep 2026 23:25:03 +0100 Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 User-Agent: Mozilla Thunderbird Subject: Re: [PATCH v19 6/7] firmware: arm_rmm: Ensure the RMM has GPT entries for memory Content-Language: en-GB To: Shanker Donthineni , kvm@vger.kernel.org, kvmarm@lists.linux.dev Cc: maz@kernel.org, will@kernel.org, catalin.marinas@arm.com, linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org, steven.price@arm.com, aneesh.kumar@kernel.org, oupton@kernel.org, gshan@redhat.com, joey.gouly@arm.com, tabba@google.com, yuzenghui@huawei.com, linux-coco@lists.linux.dev, gankulkarni@os.amperecomputing.com, alpergun@google.com, fj0570is@fujitsu.com, WeiLin.Chang@arm.com, lpieralisi@kernel.org, enju.kohei@fujitsu.com, sudeep.holla@arm.com, jonathan.cameron@oss.qualcomm.com References: <20260924135201.850038-1-suzuki.poulose@arm.com> <20260924135201.850038-7-suzuki.poulose@arm.com> <97938711-a762-4a06-a345-8e929f2e33a8@nvidia.com> From: Suzuki K Poulose In-Reply-To: <97938711-a762-4a06-a345-8e929f2e33a8@nvidia.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit On 29/09/2026 23:17, Shanker Donthineni wrote: > Hi Suzuki, > > On 9/24/2026 8:52 AM, Suzuki K Poulose wrote: >> External email: Use caution opening links or attachments >> >> >> From: Steven Price >> >> The RMM maintains the state of all the granules in the system to make >> sure that the host is abiding by the rules. This state can be maintained >> at different granularity, per page (TRACKING_FINE) or per region >> (TRACKING_COARSE or TRACKING_INTERMEDIATE). The region size depends on >> the >> underlying "RMI_GRANULE_SIZE". For a "coarse"/"intermediate" region, >> all pages in the region must be of the same state, this implies we >> need to >> have "fine" tracking for DRAM, so that we can delegate individual pages. >> >> For now we only support a statically carved out memory for tracking >> granules for the "fine" regions. This can be extended in the future to >> allow modifying the tracking granularity and remove the need for a >> static allocation by the firmware. Like the commit description says here ... >> +/* >> + * Make sure the area is tracked by RMM at FINE granularity. >> + * We do not support changing the tracking yet. >> + */ >> +static int rmi_verify_memory_tracking(phys_addr_t start, phys_addr_t >> end) >> +{ >> +       while (start < end) { >> +               unsigned long ret, category, state, next; >> + >> +               ret = rmi_granule_tracking_get(start, end, &category, >> &state, &next); >> +               if (ret != RMI_SUCCESS) >> +                       return -ENOMEM; >> + >> +               if (WARN_ON(next <= start)) >> +                       return -ENXIO; >> + >> +               if (state != RMI_TRACKING_FINE || >> +                   category != RMI_MEM_CATEGORY_CONVENTIONAL) { >> +                       /* TODO: Set granule tracking in this case */ >> +                       pr_err("Granule tracking for region isn't >> fine/conventional: %llx-%lx\n", >> +                              start, next); >> +                       return -ENODEV; > > This rejects any region the RMM does not already report as FINE, > but the initial tracking state is IMPLEMENTATION DEFINED, nothing > in DEN0137 Beta3 requires it to be FINE out of reset, and the full > RMI_TRACKING_* range is legal to observe here. Agreed. This was a deliberate decision to keep this series simple. Once we have a base support merged, we can add support for dynamic tracking and GPTs. Kind regards Suzuki > > On NVIDIA platforms it is not FINE, so this path fails and CCA cannot > be enabled at all. The TODO above is therefore load-bearing rather than > an optimisation: the driver has to promote the region with > RMI_GRANULE_TRACKING_SET and re-query, not bail out. > > -Shanker > >> +               } >> +               start = next; >> +       } >> + >> +       return 0; >> +} >> + >> +/* >> + * rmi_gpt_info - Query the GPT info for the given PAR. >> + * @start: Base of the physical address region >> + * @end: Top of the physical address region >> + * @out_top: Top of the physical address region for which >> + *             the GPT @out_gpt_par_state is valid >> + * @out_gpt_par_state: State of the GPT covered by [start, out_top) >> + */ >> +static long rmi_gpt_info(unsigned long start, unsigned long end, >> +                        unsigned long *out_top, >> +                        unsigned long *out_gpt_par_state) >> +{ >> +       struct arm_smccc_1_2_regs regs = { >> +               SMC_RMI_GPT_INFO, start, end, >> +       }; >> + >> +       rmi_smccc_invoke(®s); >> +       if (regs.a0 != RMI_SUCCESS) >> +               return regs.a0; >> + >> +       if (out_top) >> +               *out_top = regs.a1; >> +       if (out_gpt_par_state) >> +               *out_gpt_par_state = regs.a2; >> + >> +       return RMI_SUCCESS; >> +} >> + >> +/* >> + * We do not support creating L1 GPTs yet. So, make sure that >> + * all the regions are managed by the firmware. >> + */ >> +static int rmi_verify_gpt_firmware_managed(phys_addr_t start, >> phys_addr_t end) >> +{ >> +       unsigned long l0gpt_sz; >> +       unsigned long next, par_state; >> + >> +       l0gpt_sz = 1UL << (30 + FIELD_GET(RMI_FEATURE_REGISTER_1_L0GPTSZ, >> +                                         rmi_feat_reg(1))); >> +       start = ALIGN_DOWN(start, l0gpt_sz); >> +       end = ALIGN(end, l0gpt_sz); >> + >> +       while (start < end) { >> +               long ret = rmi_gpt_info(start, end, &next, &par_state); >> + >> +               if (ret != RMI_SUCCESS) >> +                       return -ENOMEM; >> + >> +               if (WARN_ON(next <= start)) >> +                       return -ENXIO; >> + >> +               if (par_state != RMI_GPT_PAR_PLAT) { >> +                       pr_err("GPT for the region is not managed by >> firmware %llx-%lx\n", >> +                               start, next); >> +                       return -ENOMEM; >> +               } >> +               start = next; >> +       } >> + >> +       return 0; >> +} >> + >> +static int rmi_prepare_memory(phys_addr_t start, phys_addr_t end) >> +{ >> +       int ret; >> + >> +       if (start >= end) >> +               return -EINVAL; >> + >> +       ret = rmi_verify_memory_tracking(start, end); >> +       if (ret) >> +               return ret; >> + >> +       return rmi_verify_gpt_firmware_managed(start, end); >> +} >> + >> +static int rmi_init_metadata(void) >> +{ >> +       phys_addr_t start, end; >> +       struct memblock_region *r; >> + >> +       for_each_mem_region(r) { >> +               int ret; >> + >> +               /* Firmware-reserved NOMAP regions are not usable >> system RAM */ >> +               if (memblock_is_nomap(r)) >> +                       continue; >> + >> +               start = PAGE_ALIGN(r->base); >> +               end = PAGE_ALIGN_DOWN(r->base + r->size); >> +               /* Too small ? */ >> +               if (start >= end) >> +                       continue; >> + >> +               ret = rmi_prepare_memory(start, end); >> +               if (ret) >> +                       return ret; >> +       } >> + >> +       return 0; >> +} >> + >> +static int rmi_memory_notifier(struct notifier_block *nb, >> +                              unsigned long action, void *data) >> +{ >> +       struct memory_notify *arg = data; >> +       phys_addr_t start, end; >> +       int ret; >> + >> +       if (action != MEM_GOING_ONLINE) >> +               return NOTIFY_DONE; >> + >> +       start = PFN_PHYS(arg->start_pfn); >> +       end = PFN_PHYS(arg->start_pfn + arg->nr_pages); >> +       ret = rmi_prepare_memory(start, end); >> + >> +       return notifier_from_errno(ret); >> +} >> + >> +static struct notifier_block rmi_memory_nb = { >> +       .notifier_call = rmi_memory_notifier, >> +}; >> + >> +bool is_rmi_available(void) >> +{ >> +       return arm64_rmi_is_available; >> +} >> +EXPORT_SYMBOL_GPL(is_rmi_available); >> + >> +static int rmi_init_memory(void) >> +{ >> +       int ret; >> + >> +       ret = rmi_init_metadata(); >> +       if (ret) >> +               return ret; >> + >> +       return register_memory_notifier(&rmi_memory_nb); >> +} >> + >>   static int __init arm64_init_rmi(void) >>   { >>          int ret; >> @@ -843,9 +1049,19 @@ static int __init arm64_init_rmi(void) >>          if (ret) { >>                  pr_err("RMM activate failed (%d)\n", ret); >>                  ret = ret < 0 ? ret : -ENXIO; >> +               return ret; >>          } >> >> -       return ret; >> +       ret = rmi_init_memory(); >> +       if (ret) { >> +               /* Deactivate the RMM */ >> +               WARN_ON(rmi_sro_memxfer_cmd(sro, GFP_KERNEL, >> SMC_RMI_RMM_DEACTIVATE)); >> +               return ret; >> +       } >> + >> +       arm64_rmi_is_available = true; >> +       pr_info("RMI configured\n"); >> +       return 0; >>   } >> >>   /* >> diff --git a/include/linux/arm-rmi-cmds.h b/include/linux/arm-rmi-cmds.h >> index b03974fd8168c..5b177bb176326 100644 >> --- a/include/linux/arm-rmi-cmds.h >> +++ b/include/linux/arm-rmi-cmds.h >> @@ -70,6 +70,8 @@ static inline int rmi_undelegate_page(phys_addr_t phys) >>          return rmi_undelegate_range(phys, PAGE_SIZE); >>   } >> >> +bool is_rmi_available(void); >> + >>   long rmi_sro_memxfer_execute(struct rmi_sro_state *sro, gfp_t gfp); >>   void rmi_sro_free(struct rmi_sro_state *sro); >>   long rmi_sro_execute(struct arm_smccc_1_2_regs *regs); >> -- >> 2.43.0 >> >