From: Jan Beulich <jbeulich@suse.com>
To: Wei Chen <Wei.Chen@arm.com>
Cc: Bertrand Marquis <Bertrand.Marquis@arm.com>,
"xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>,
"sstabellini@kernel.org" <sstabellini@kernel.org>,
"julien@xen.org" <julien@xen.org>
Subject: Re: [PATCH 04/37] xen: introduce an arch helper for default dma zone status
Date: Tue, 18 Jan 2022 09:16:08 +0100 [thread overview]
Message-ID: <378fcc9b-abfe-28f2-ee67-f5510e99436c@suse.com> (raw)
In-Reply-To: <PAXPR08MB7420CD3B7CFB4D2D81019F479E589@PAXPR08MB7420.eurprd08.prod.outlook.com>
On 18.01.2022 08:51, Wei Chen wrote:
>> -----Original Message-----
>> From: Jan Beulich <jbeulich@suse.com>
>> Sent: 2022年1月18日 0:11
>> On 23.09.2021 14:02, Wei Chen wrote:
>>> In current code, when Xen is running in a multiple nodes NUMA
>>> system, it will set dma_bitsize in end_boot_allocator to reserve
>>> some low address memory for DMA.
>>>
>>> There are some x86 implications in current implementation. Becuase
>>> on x86, memory starts from 0. On a multiple nodes NUMA system, if
>>> a single node contains the majority or all of the DMA memory. x86
>>> prefer to give out memory from non-local allocations rather than
>>> exhausting the DMA memory ranges. Hence x86 use dma_bitsize to set
>>> aside some largely arbitrary amount memory for DMA memory ranges.
>>> The allocations from these memory ranges would happen only after
>>> exhausting all other nodes' memory.
>>>
>>> But the implications are not shared across all architectures. For
>>> example, Arm doesn't have these implications. So in this patch, we
>>> introduce an arch_have_default_dmazone helper for arch to determine
>>> that it need to set dma_bitsize for reserve DMA allocations or not.
>>
>> How would Arm guarantee availability of memory below a certain
>> boundary for limited-capability devices? Or is there no need
>> because there's an assumption that I/O for such devices would
>> always pass through an IOMMU, lifting address size restrictions?
>> (I guess in a !PV build on x86 we could also get rid of such a
>> reservation.)
>
> On Arm, we still can have some devices with limited DMA capability.
> And we also don't force all such devices to use IOMMU. This devices
> will affect the dma_bitsize. Like RPi platform, it sets its dma_bitsize
> to 30. But in multiple NUMA nodes system, Arm doesn't have a default
> DMA zone. Multiple nodes is not a constraint on dma_bitsize. And some
> previous discussions are placed here [1].
I'm afraid that doesn't give me more clues. For example, in the mail
being replied to there I find "That means, only first 4GB memory can
be used for DMA." Yet that's not an implication from setting
dma_bitsize. DMA is fine to occur to any address. The special address
range is being held back in case in particular Dom0 is in need of such
a range to perform I/O to _some_ devices.
>>> --- a/xen/arch/x86/numa.c
>>> +++ b/xen/arch/x86/numa.c
>>> @@ -371,6 +371,11 @@ unsigned int __init arch_get_dma_bitsize(void)
>>> + PAGE_SHIFT, 32);
>>> }
>>>
>>> +unsigned int arch_have_default_dmazone(void)
>>> +{
>>> + return ( num_online_nodes() > 1 ) ? 1 : 0;
>>> +}
>>
>> According to the expression and ...
>>
>>> --- a/xen/common/page_alloc.c
>>> +++ b/xen/common/page_alloc.c
>>> @@ -1889,7 +1889,7 @@ void __init end_boot_allocator(void)
>>> }
>>> nr_bootmem_regions = 0;
>>>
>>> - if ( !dma_bitsize && (num_online_nodes() > 1) )
>>> + if ( !dma_bitsize && arch_have_default_dmazone() )
>>> dma_bitsize = arch_get_dma_bitsize();
>>
>> ... the use site, you mean the function to return boolean. Please
>> indicate so by making it have a return type of "bool". Independent
>> of that you don't need a conditional expression above, nor
>> (malformed) use of parentheses. I further wonder whether ...
>>
>
> I will fix them in next version. But I am not very clear about
> this comment "of that you don't need a conditional expression above",
> The "above" indicates this line:
> "return ( num_online_nodes() > 1 ) ? 1 : 0;"?
Yes. Even without the use of bool such an expression is a more
complicated form of
return num_online_nodes() > 1;
where we'd prefer to use the simpler variant for being easier to
read / follow.
Jan
> [1] https://lists.xenproject.org/archives/html/xen-devel/2021-08/msg00772.html
>
>
next prev parent reply other threads:[~2022-01-18 8:16 UTC|newest]
Thread overview: 192+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-23 12:01 [PATCH 00/37] Add device tree based NUMA support to Arm Wei Chen
2021-09-23 12:02 ` [PATCH 01/37] xen/arm: Print a 64-bit number in hex from early uart Wei Chen
2021-09-23 12:02 ` [PATCH 02/37] xen: introduce a Kconfig option to configure NUMA nodes number Wei Chen
2021-09-23 23:45 ` Stefano Stabellini
2021-09-24 1:24 ` Wei Chen
2021-09-24 8:55 ` Jan Beulich
2021-09-24 10:33 ` Wei Chen
2021-09-24 10:47 ` Jan Beulich
2021-09-23 12:02 ` [PATCH 03/37] xen/x86: Initialize memnodemapsize while faking NUMA node Wei Chen
2021-09-24 8:57 ` Jan Beulich
2021-09-24 10:34 ` Wei Chen
2021-09-23 12:02 ` [PATCH 04/37] xen: introduce an arch helper for default dma zone status Wei Chen
2021-09-23 23:55 ` Stefano Stabellini
2021-09-24 1:50 ` Wei Chen
2022-01-17 16:10 ` Jan Beulich
2022-01-18 7:51 ` Wei Chen
2022-01-18 8:16 ` Jan Beulich [this message]
2022-01-18 9:20 ` Wei Chen
2022-01-18 14:16 ` Jan Beulich
2022-01-19 2:49 ` Wei Chen
2022-01-19 7:50 ` Jan Beulich
2022-01-19 8:33 ` Wei Chen
2021-09-23 12:02 ` [PATCH 05/37] xen: decouple NUMA from ACPI in Kconfig Wei Chen
2021-09-23 12:02 ` [PATCH 06/37] xen/arm: use !CONFIG_NUMA to keep fake NUMA API Wei Chen
2021-09-24 0:05 ` Stefano Stabellini
2021-09-24 10:21 ` Wei Chen
2021-09-23 12:02 ` [PATCH 07/37] xen/x86: use paddr_t for addresses in NUMA node structure Wei Chen
2021-09-24 0:11 ` Stefano Stabellini
2021-09-24 0:13 ` Stefano Stabellini
2021-09-24 3:00 ` Wei Chen
2022-01-18 15:22 ` Jan Beulich
2022-01-19 6:33 ` Wei Chen
2022-01-19 7:55 ` Jan Beulich
2022-01-19 8:36 ` Wei Chen
2021-09-23 12:02 ` [PATCH 08/37] xen/x86: add detection of discontinous node memory range Wei Chen
2021-09-24 0:25 ` Stefano Stabellini
2021-09-24 4:28 ` Wei Chen
2021-09-24 19:52 ` Stefano Stabellini
2021-09-26 10:11 ` Wei Chen
2021-09-27 3:13 ` Stefano Stabellini
2021-09-27 5:05 ` Stefano Stabellini
2021-09-27 9:50 ` Wei Chen
2021-09-27 17:19 ` Stefano Stabellini
2021-09-28 4:41 ` Wei Chen
2021-09-28 4:59 ` Stefano Stabellini
2022-01-18 16:13 ` Jan Beulich
2022-01-19 7:33 ` Wei Chen
2022-01-19 8:01 ` Jan Beulich
2022-01-19 8:24 ` Wei Chen
2021-09-23 12:02 ` [PATCH 09/37] xen/x86: introduce two helpers to access memory hotplug end Wei Chen
2021-09-24 0:29 ` Stefano Stabellini
2021-09-24 4:21 ` Wei Chen
2022-01-24 16:24 ` Jan Beulich
2022-01-26 7:53 ` Wei Chen
2021-09-23 12:02 ` [PATCH 10/37] xen/x86: use helpers to access/update mem_hotplug Wei Chen
2021-09-24 0:31 ` Stefano Stabellini
2021-09-24 4:29 ` Wei Chen
2022-01-24 16:29 ` Jan Beulich
2022-01-26 7:58 ` Wei Chen
2021-09-23 12:02 ` [PATCH 11/37] xen/x86: abstract neutral code from acpi_numa_memory_affinity_init Wei Chen
2021-09-24 0:38 ` Stefano Stabellini
2022-01-24 16:50 ` Jan Beulich
2022-01-26 10:39 ` Wei Chen
2021-09-23 12:02 ` [PATCH 12/37] xen/x86: decouple nodes_cover_memory from E820 map Wei Chen
2021-09-24 0:39 ` Stefano Stabellini
2022-01-24 16:59 ` Jan Beulich
2022-01-27 8:03 ` Wei Chen
2022-01-27 8:08 ` Jan Beulich
2022-01-27 9:03 ` Wei Chen
2022-01-27 9:22 ` Jan Beulich
2022-01-27 9:27 ` Wei Chen
2021-09-23 12:02 ` [PATCH 13/37] xen/x86: decouple processor_nodes_parsed from acpi numa functions Wei Chen
2021-09-24 0:40 ` Stefano Stabellini
2022-01-25 9:49 ` Jan Beulich
2022-01-27 8:06 ` Wei Chen
2021-09-23 12:02 ` [PATCH 14/37] xen/x86: use name fw_numa to replace acpi_numa Wei Chen
2021-09-24 0:40 ` Stefano Stabellini
2022-01-25 10:12 ` Jan Beulich
2022-01-27 8:09 ` Wei Chen
2021-09-23 12:02 ` [PATCH 15/37] xen/x86: rename acpi_scan_nodes to numa_scan_nodes Wei Chen
2021-09-24 0:40 ` Stefano Stabellini
2022-01-25 10:17 ` Jan Beulich
2022-01-27 8:14 ` Wei Chen
2021-09-23 12:02 ` [PATCH 16/37] xen/x86: export srat_bad to external Wei Chen
2021-09-24 0:41 ` Stefano Stabellini
2022-01-25 10:22 ` Jan Beulich
2022-01-27 8:35 ` Wei Chen
2022-01-27 8:37 ` Jan Beulich
2022-01-27 8:47 ` Wei Chen
2021-09-23 12:02 ` [PATCH 17/37] xen/x86: use CONFIG_NUMA to gate numa_scan_nodes Wei Chen
2021-09-24 0:41 ` Stefano Stabellini
2022-01-25 10:26 ` Jan Beulich
2022-01-27 8:37 ` Wei Chen
2021-09-23 12:02 ` [PATCH 18/37] xen: move NUMA common code from x86 to common Wei Chen
2021-09-23 12:02 ` [PATCH 19/37] xen/x86: promote VIRTUAL_BUG_ON to ASSERT in Wei Chen
2022-01-17 16:21 ` Jan Beulich
2022-01-18 7:52 ` Wei Chen
2021-09-23 12:02 ` [PATCH 20/37] xen: introduce CONFIG_EFI to stub API for non-EFI architecture Wei Chen
2021-09-24 1:15 ` Stefano Stabellini
2021-09-24 4:34 ` Wei Chen
2021-09-24 7:58 ` Jan Beulich
2021-09-24 10:31 ` Wei Chen
2021-09-24 10:49 ` Jan Beulich
2021-09-26 10:25 ` Wei Chen
2021-09-27 10:28 ` Wei Chen
2021-09-28 0:59 ` Stefano Stabellini
2021-09-28 4:16 ` Wei Chen
2021-09-28 5:01 ` Stefano Stabellini
2021-09-28 8:02 ` Jan Beulich
2021-10-03 23:28 ` Wei Chen
2022-01-25 10:34 ` Jan Beulich
2022-01-27 8:44 ` Wei Chen
2022-01-27 8:51 ` Wei Chen
2022-01-27 9:00 ` Jan Beulich
2022-01-27 9:09 ` Wei Chen
2022-01-27 9:16 ` Jan Beulich
2022-01-27 9:25 ` Wei Chen
2022-01-27 9:27 ` Jan Beulich
2022-01-27 10:00 ` Julien Grall
2022-01-28 4:35 ` Wei Chen
2021-09-23 12:02 ` [PATCH 21/37] xen/arm: Keep memory nodes in dtb for NUMA when boot from EFI Wei Chen
2021-09-24 1:23 ` Stefano Stabellini
2021-09-24 4:36 ` Wei Chen
2022-01-25 10:38 ` Jan Beulich
2022-01-27 8:45 ` Wei Chen
2021-09-23 12:02 ` [PATCH 22/37] xen/arm: use NR_MEM_BANKS to override default NR_NODE_MEMBLKS Wei Chen
2021-09-24 1:34 ` Stefano Stabellini
2021-09-26 13:13 ` Wei Chen
2021-09-27 3:25 ` Stefano Stabellini
2021-09-27 4:18 ` Wei Chen
2021-09-27 4:59 ` Stefano Stabellini
2021-09-27 6:25 ` Julien Grall
2021-09-27 6:46 ` Wei Chen
2021-09-27 6:53 ` Wei Chen
2021-09-27 7:35 ` Julien Grall
2021-09-27 10:21 ` Wei Chen
2021-09-27 10:39 ` Julien Grall
2021-09-27 16:58 ` Stefano Stabellini
2021-09-28 2:57 ` Wei Chen
2021-09-23 12:02 ` [PATCH 23/37] xen/arm: implement node distance helpers for Arm Wei Chen
2021-09-24 1:46 ` Stefano Stabellini
2021-09-24 4:41 ` Wei Chen
2021-09-24 19:36 ` Stefano Stabellini
2021-09-26 10:15 ` Wei Chen
2021-09-23 12:02 ` [PATCH 24/37] xen/arm: implement two arch helpers to get memory map info Wei Chen
2021-09-24 2:06 ` Stefano Stabellini
2021-09-24 4:42 ` Wei Chen
2021-09-23 12:02 ` [PATCH 25/37] xen/arm: implement bad_srat for Arm NUMA initialization Wei Chen
2021-09-24 2:09 ` Stefano Stabellini
2021-09-24 4:45 ` Wei Chen
2021-09-24 8:07 ` Jan Beulich
2021-09-24 19:33 ` Stefano Stabellini
2021-09-23 12:02 ` [PATCH 26/37] xen/arm: build NUMA cpu_to_node map in dt_smp_init_cpus Wei Chen
2021-09-24 2:26 ` Stefano Stabellini
2021-09-24 4:25 ` Wei Chen
2021-09-23 12:02 ` [PATCH 27/37] xen/arm: Add boot and secondary CPU to NUMA system Wei Chen
2021-09-23 12:02 ` [PATCH 28/37] xen/arm: stub memory hotplug access helpers for Arm Wei Chen
2021-09-24 2:33 ` Stefano Stabellini
2021-09-24 4:26 ` Wei Chen
2021-09-23 12:02 ` [PATCH 29/37] xen/arm: introduce a helper to parse device tree processor node Wei Chen
2021-09-24 2:44 ` Stefano Stabellini
2021-09-24 4:46 ` Wei Chen
2021-09-23 12:02 ` [PATCH 30/37] xen/arm: introduce a helper to parse device tree memory node Wei Chen
2021-09-24 3:05 ` Stefano Stabellini
2021-09-24 7:54 ` Wei Chen
2021-09-23 12:02 ` [PATCH 31/37] xen/arm: introduce a helper to parse device tree NUMA distance map Wei Chen
2021-09-24 3:05 ` Stefano Stabellini
2021-09-24 5:23 ` Wei Chen
2021-09-23 12:02 ` [PATCH 32/37] xen/arm: unified entry to parse all NUMA data from device tree Wei Chen
2021-09-24 3:16 ` Stefano Stabellini
2021-09-24 7:58 ` Wei Chen
2021-09-24 19:42 ` Stefano Stabellini
2021-09-23 12:02 ` [PATCH 33/37] xen/arm: keep guest still be NUMA unware Wei Chen
2021-09-24 3:19 ` Stefano Stabellini
2021-09-24 10:23 ` Wei Chen
2021-09-23 12:02 ` [PATCH 34/37] xen/arm: enable device tree based NUMA in system init Wei Chen
2021-09-24 3:28 ` Stefano Stabellini
2021-09-24 9:52 ` Wei Chen
2021-09-23 12:02 ` [PATCH 35/37] xen/arm: use CONFIG_NUMA to gate node_online_map in smpboot Wei Chen
2021-09-23 12:02 ` [PATCH 36/37] xen/arm: Provide Kconfig options for Arm to enable NUMA Wei Chen
2021-09-24 3:31 ` Stefano Stabellini
2021-09-24 10:13 ` Wei Chen
2021-09-24 19:39 ` Stefano Stabellini
2021-09-27 8:33 ` Jan Beulich
2021-09-27 8:45 ` Julien Grall
2021-09-27 9:17 ` Jan Beulich
2021-09-27 17:17 ` Stefano Stabellini
2021-09-28 2:59 ` Wei Chen
2021-09-28 3:30 ` Stefano Stabellini
2021-09-24 10:25 ` Jan Beulich
2021-09-24 10:37 ` Wei Chen
2021-09-23 12:02 ` [PATCH 37/37] docs: update numa command line to support Arm Wei Chen
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=378fcc9b-abfe-28f2-ee67-f5510e99436c@suse.com \
--to=jbeulich@suse.com \
--cc=Bertrand.Marquis@arm.com \
--cc=Wei.Chen@arm.com \
--cc=julien@xen.org \
--cc=sstabellini@kernel.org \
--cc=xen-devel@lists.xenproject.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).