Message-ID: <028da60d-4f67-60bf-4360-83fe37bcbb7e@linuxfoundation.org>
Date:   Fri, 9 Apr 2021 14:19:57 -0600
From:   Shuah Khan <skhan@...uxfoundation.org>
To:     Suravee Suthikulpanit <suravee.suthikulpanit@....com>,
        linux-kernel@...r.kernel.org, iommu@...ts.linux-foundation.org
Cc:     joro@...tes.org, will@...nel.org, jsnitsel@...hat.com,
        pmenzel@...gen.mpg.de, Jon.Grimm@....com,
        Tj <ml.linux@...oe.vision>,
        Alexander Monakov <amonakov@...ras.ru>,
        David Coe <david.coe@...e.co.uk>,
        Shuah Khan <skhan@...uxfoundation.org>,
        1917203@...s.launchpad.net
Subject: Re: [PATCH 2/2] iommu/amd: Remove performance counter
 pre-initialization test

On 4/9/21 2:00 PM, Shuah Khan wrote:
> On 4/9/21 2:58 AM, Suravee Suthikulpanit wrote:
>> In early AMD desktop/mobile platforms (during 2013), when the IOMMU
>> Performance Counter (PMC) support was first introduced in
>> commit 30861ddc9cca ("perf/x86/amd: Add IOMMU Performance Counter
>> resource management"), there was a HW bug where the counters could not
>> be accessed. The result was reading of the counter always return zero.
>>
>> At the time, the suggested workaround was to add a test logic prior
>> to initializing the PMC feature to check if the counters can be 
>> programmed
>> and read back the same value. This has been working fine until the more
>> recent desktop/mobile platforms start enabling power gating for the PMC,
>> which prevents access to the counters. This results in the PMC support
>> being disabled unnecesarily.
>>
>> Unfortunatly, there is no documentation of since which generation
>> of hardware the original PMC HW bug was fixed. Although, it was fixed
>> soon after the first introduction of the PMC. Base on this, we assume
>> that the buggy platforms are less likely to be in used, and it should
>> be relatively safe to remove this legacy logic.
>>
>> Link: 
>> https://lore.kernel.org/linux-iommu/alpine.LNX.3.20.13.2006030935570.3181@monopod.intra.ispras.ru/ 
>>
>> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=201753
>> Cc: Tj (Elloe Linux) <ml.linux@...oe.vision>
>> Cc: Shuah Khan <skhan@...uxfoundation.org>
>> Cc: Alexander Monakov <amonakov@...ras.ru>
>> Cc: David Coe <david.coe@...e.co.uk>
>> Cc: Paul Menzel <pmenzel@...gen.mpg.de>
>> Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@....com>
>> ---
> 
> 
> Tested-by: Shuah Khan <skhan@...uxfoundation.org>
> 

Revert + this patch - same as my test on Ryzen 5

On AMD Ryzen 7 4700G with Radeon Graphics

These look real odd to me. Let me know if I should look further.

sudo ./perf stat -e 'amd_iommu_0/cmd_processed/, 
amd_iommu_0/cmd_processed_inv/, amd_iommu_0/ign_rd_wr_mmio_1ff8h/, 
amd_iommu_0/int_dte_hit/, amd_iommu_0/int_dte_mis/, 
amd_iommu_0/mem_dte_hit/, amd_iommu_0/mem_dte_mis/, 
amd_iommu_0/mem_iommu_tlb_pde_hit/, amd_iommu_0/mem_iommu_tlb_pde_mis/, 
amd_iommu_0/mem_iommu_tlb_pte_hit/, amd_iommu_0/mem_iommu_tlb_pte_mis/, 
amd_iommu_0/mem_pass_excl/, amd_iommu_0/mem_pass_pretrans/, 
amd_iommu_0/mem_pass_untrans/, amd_iommu_0/mem_target_abort/, 
amd_iommu_0/mem_trans_total/, amd_iommu_0/page_tbl_read_gst/, 
amd_iommu_0/page_tbl_read_nst/, amd_iommu_0/page_tbl_read_tot/, 
amd_iommu_0/smi_blk/, amd_iommu_0/smi_recv/, amd_iommu_0/tlb_inv/, 
amd_iommu_0/vapic_int_guest/, amd_iommu_0/vapic_int_non_guest/' sleep 10

  Performance counter stats for 'system wide':

17,761,952,514,865,374      amd_iommu_0/cmd_processed/ 
                    (33.28%)
18,582,155,570,607,472       amd_iommu_0/cmd_processed_inv/ 
                         (33.32%)
                  0       amd_iommu_0/ign_rd_wr_mmio_1ff8h/ 
                         (33.36%)
5,056,087,645,262,255       amd_iommu_0/int_dte_hit/ 
                  (33.40%)
32,831,106,446,308,888       amd_iommu_0/int_dte_mis/ 
                   (33.44%)
13,461,819,655,591,296       amd_iommu_0/mem_dte_hit/ 
                   (33.45%)
208,555,436,221,050,464       amd_iommu_0/mem_dte_mis/ 
                    (33.47%)
196,824,154,635,609,888       amd_iommu_0/mem_iommu_tlb_pde_hit/ 
                              (33.46%)
193,552,630,440,410,144       amd_iommu_0/mem_iommu_tlb_pde_mis/ 
                              (33.45%)
176,936,647,809,098,368       amd_iommu_0/mem_iommu_tlb_pte_hit/ 
                              (33.41%)
184,737,401,623,626,464       amd_iommu_0/mem_iommu_tlb_pte_mis/ 
                              (33.37%)
                  0       amd_iommu_0/mem_pass_excl/ 
                  (33.33%)
                  0       amd_iommu_0/mem_pass_pretrans/ 
                      (33.30%)
                  0       amd_iommu_0/mem_pass_untrans/ 
                     (33.28%)
                  0       amd_iommu_0/mem_target_abort/ 
                     (33.27%)
245,383,212,924,004,288       amd_iommu_0/mem_trans_total/ 
                        (33.27%)
                  0       amd_iommu_0/page_tbl_read_gst/ 
                      (33.28%)
262,267,045,917,967,264       amd_iommu_0/page_tbl_read_nst/ 
                          (33.27%)
256,308,216,913,137,600       amd_iommu_0/page_tbl_read_tot/ 
                          (33.28%)
                  0       amd_iommu_0/smi_blk/ 
                (33.27%)
                  0       amd_iommu_0/smi_recv/ 
                (33.27%)
                  0       amd_iommu_0/tlb_inv/ 
                (33.27%)
                  0       amd_iommu_0/vapic_int_guest/ 
                    (33.26%)
38,913,544,420,579,888       amd_iommu_0/vapic_int_non_guest/ 
                           (33.27%)

       10.003967760 seconds time elapsed

thanks,
-- Shuah

Powered by blists - more mailing lists