| Message ID | 20200926041928.9xJHGgkah%akpm@linux-foundation.org |
|---|---|
| State | New |
| Headers | show |
| Series | [1/9] mm, THP, swap: fix allocating cluster for swapfile by mistake | expand |
The explanations here do not make sense. On Fri, Sep 25, 2020 at 9:19 PM Andrew Morton <akpm@linux-foundation.org> wrote: > > > There are 2 issues here: > > a. The sysfs memory and node's layouts are broken due to these multiple > links > > b. The link errors in link_mem_sections() should not lead to a system > panic. > > To address a. register_mem_sect_under_node should not rely on the system > state to detect whether the link operation is triggered by a hot plug > operation or not. This is addressed by the patches 1 and 2 of this series. > > The patch 3 is addressing the point b. > > This patch (of 2): > > The memmap_context enum is used to detect whether a memory operation is due > to a hot-add operation or happening at boot time. > > Make it general to the hotplug operation and rename it as meminit_context. > > There is no functional change introduced by this patch So far so good. But there is no "patch 3" that addresses point (b) in this series. I see it on lore, but it's not part of what actually got sent to me, so the commit message for patch 1 now makes no sense any more. Linus
On Sat 26-09-20 10:32:15, Linus Torvalds wrote: > The explanations here do not make sense. > > On Fri, Sep 25, 2020 at 9:19 PM Andrew Morton <akpm@linux-foundation.org> wrote: > > > > > > There are 2 issues here: > > > > a. The sysfs memory and node's layouts are broken due to these multiple > > links > > > > b. The link errors in link_mem_sections() should not lead to a system > > panic. > > > > To address a. register_mem_sect_under_node should not rely on the system > > state to detect whether the link operation is triggered by a hot plug > > operation or not. This is addressed by the patches 1 and 2 of this series. > > > > The patch 3 is addressing the point b. > > > > This patch (of 2): > > > > The memmap_context enum is used to detect whether a memory operation is due > > to a hot-add operation or happening at boot time. > > > > Make it general to the hotplug operation and rename it as meminit_context. > > > > There is no functional change introduced by this patch > > So far so good. > > But there is no "patch 3" that addresses point (b) in this series. > > I see it on lore, but it's not part of what actually got sent to me, > so the commit message for patch 1 now makes no sense any more. You are right. There should be patch 2 posted as well. And I would argue that the 3rd patch should go in with them as well. This is a preparatory patch only.
Le 26/09/2020 à 19:32, Linus Torvalds a écrit : > The explanations here do not make sense. > > On Fri, Sep 25, 2020 at 9:19 PM Andrew Morton <akpm@linux-foundation.org> wrote: >> >> >> There are 2 issues here: >> >> a. The sysfs memory and node's layouts are broken due to these multiple >> links >> >> b. The link errors in link_mem_sections() should not lead to a system >> panic. >> >> To address a. register_mem_sect_under_node should not rely on the system >> state to detect whether the link operation is triggered by a hot plug >> operation or not. This is addressed by the patches 1 and 2 of this series. >> >> The patch 3 is addressing the point b. >> >> This patch (of 2): >> >> The memmap_context enum is used to detect whether a memory operation is due >> to a hot-add operation or happening at boot time. >> >> Make it general to the hotplug operation and rename it as meminit_context. >> >> There is no functional change introduced by this patch > > So far so good. > > But there is no "patch 3" that addresses point (b) in this series. > > I see it on lore, but it's not part of what actually got sent to me, > so the commit message for patch 1 now makes no sense any more. I agree, the commit description is a bit confusing now. I think the Andrew's idea was to queue up the 2 first patches which are cc:stable to the 5.9-rcX and later the third one which is not really fixing an issue. The BUG_ON is unlikely to be triggered once the 2 first patches are applied. Thanks, Laurent.
On Sat, 26 Sep 2020 10:32:15 -0700 Linus Torvalds <torvalds@linux-foundation.org> wrote: > The explanations here do not make sense. > > On Fri, Sep 25, 2020 at 9:19 PM Andrew Morton <akpm@linux-foundation.org> wrote: > > > > > > There are 2 issues here: > > > > a. The sysfs memory and node's layouts are broken due to these multiple > > links > > > > b. The link errors in link_mem_sections() should not lead to a system > > panic. > > > > To address a. register_mem_sect_under_node should not rely on the system > > state to detect whether the link operation is triggered by a hot plug > > operation or not. This is addressed by the patches 1 and 2 of this series. > > > > The patch 3 is addressing the point b. > > > > This patch (of 2): > > > > The memmap_context enum is used to detect whether a memory operation is due > > to a hot-add operation or happening at boot time. > > > > Make it general to the hotplug operation and rename it as meminit_context. > > > > There is no functional change introduced by this patch > > So far so good. > > But there is no "patch 3" that addresses point (b) in this series. > > I see it on lore, but it's not part of what actually got sent to me, > so the commit message for patch 1 now makes no sense any more. > 1/3 and 2/3 were cc:stable and 3/3 was not. As far as I can tell, 3/3 is rather theoretical once 2/3 has done its work, so I held it off for the next merge window.
On Tue, Sep 29, 2020 at 1:37 PM Andrew Morton <akpm@linux-foundation.org> wrote: > > 1/3 and 2/3 were cc:stable and 3/3 was not. As far as I can tell, 3/3 > is rather theoretical once 2/3 has done its work, so I held it off for > the next merge window. That's not the problem, holding off is fine. The problem is that the commit messages are garbage as a result. They were written as a series of three, but the patches weren't _sent_ as a series of three. So if you split up a series like that, you should look at the commit mesages and edit them appropriately. Or, in fact, try to avoid such commit messages that depend on other commit messages entirely. Linus
On Wed 30-09-20 09:00:26, Linus Torvalds wrote: > On Tue, Sep 29, 2020 at 1:37 PM Andrew Morton <akpm@linux-foundation.org> wrote: > > > > 1/3 and 2/3 were cc:stable and 3/3 was not. As far as I can tell, 3/3 > > is rather theoretical once 2/3 has done its work, so I held it off for > > the next merge window. > > That's not the problem, holding off is fine. > > The problem is that the commit messages are garbage as a result. They > were written as a series of three, but the patches weren't _sent_ as a > series of three. This part of the commit message is coming from a cover letter and it is something that would usually go to a merge commit (from a pull request). With Andrew's worklflow he preserves the information in the first patch of the series. It is great to have that information around because there is usually more background and a high level description. > So if you split up a series like that, you should look at the commit > mesages and edit them appropriately. Yeah, in cases like that it is better to just drop that cover letter part. I would still argue that all three could have gone in together. The last patch has not been marked for stable but it is a useful patch on its own as it drops a really distasteful BUG() on a failure mode. IMHO it wouldn't be a big deal to postpone sending these during the merge window as these patches are not addressing a regression for this part merge window. Just my 2c
--- a/arch/ia64/mm/init.c~mm-replace-memmap_context-by-meminit_context +++ a/arch/ia64/mm/init.c @@ -538,7 +538,7 @@ virtual_memmap_init(u64 start, u64 end, if (map_start < map_end) memmap_init_zone((unsigned long)(map_end - map_start), args->nid, args->zone, page_to_pfn(map_start), - MEMMAP_EARLY, NULL); + MEMINIT_EARLY, NULL); return 0; } @@ -547,8 +547,8 @@ memmap_init (unsigned long size, int nid unsigned long start_pfn) { if (!vmem_map) { - memmap_init_zone(size, nid, zone, start_pfn, MEMMAP_EARLY, - NULL); + memmap_init_zone(size, nid, zone, start_pfn, + MEMINIT_EARLY, NULL); } else { struct page *start; struct memmap_init_callback_data args; --- a/include/linux/mm.h~mm-replace-memmap_context-by-meminit_context +++ a/include/linux/mm.h @@ -2416,7 +2416,7 @@ extern int __meminit __early_pfn_to_nid( extern void set_dma_reserve(unsigned long new_dma_reserve); extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long, - enum memmap_context, struct vmem_altmap *); + enum meminit_context, struct vmem_altmap *); extern void setup_per_zone_wmarks(void); extern int __meminit init_per_zone_wmark_min(void); extern void mem_init(void); --- a/include/linux/mmzone.h~mm-replace-memmap_context-by-meminit_context +++ a/include/linux/mmzone.h @@ -824,10 +824,15 @@ bool zone_watermark_ok(struct zone *z, u unsigned int alloc_flags); bool zone_watermark_ok_safe(struct zone *z, unsigned int order, unsigned long mark, int highest_zoneidx); -enum memmap_context { - MEMMAP_EARLY, - MEMMAP_HOTPLUG, +/* + * Memory initialization context, use to differentiate memory added by + * the platform statically or via memory hotplug interface. + */ +enum meminit_context { + MEMINIT_EARLY, + MEMINIT_HOTPLUG, }; + extern void init_currently_empty_zone(struct zone *zone, unsigned long start_pfn, unsigned long size); --- a/mm/memory_hotplug.c~mm-replace-memmap_context-by-meminit_context +++ a/mm/memory_hotplug.c @@ -729,7 +729,7 @@ void __ref move_pfn_range_to_zone(struct * are reserved so nobody should be touching them so we should be safe */ memmap_init_zone(nr_pages, nid, zone_idx(zone), start_pfn, - MEMMAP_HOTPLUG, altmap); + MEMINIT_HOTPLUG, altmap); set_zone_contiguous(zone); } --- a/mm/page_alloc.c~mm-replace-memmap_context-by-meminit_context +++ a/mm/page_alloc.c @@ -5975,7 +5975,7 @@ overlap_memmap_init(unsigned long zone, * done. Non-atomic initialization, single-pass. */ void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone, - unsigned long start_pfn, enum memmap_context context, + unsigned long start_pfn, enum meminit_context context, struct vmem_altmap *altmap) { unsigned long pfn, end_pfn = start_pfn + size; @@ -6007,7 +6007,7 @@ void __meminit memmap_init_zone(unsigned * There can be holes in boot-time mem_map[]s handed to this * function. They do not exist on hotplugged memory. */ - if (context == MEMMAP_EARLY) { + if (context == MEMINIT_EARLY) { if (overlap_memmap_init(zone, &pfn)) continue; if (defer_init(nid, pfn, end_pfn)) @@ -6016,7 +6016,7 @@ void __meminit memmap_init_zone(unsigned page = pfn_to_page(pfn); __init_single_page(page, pfn, zone, nid); - if (context == MEMMAP_HOTPLUG) + if (context == MEMINIT_HOTPLUG) __SetPageReserved(page); /* @@ -6099,7 +6099,7 @@ void __ref memmap_init_zone_device(struc * check here not to call set_pageblock_migratetype() against * pfn out of zone. * - * Please note that MEMMAP_HOTPLUG path doesn't clear memmap + * Please note that MEMINIT_HOTPLUG path doesn't clear memmap * because this is done early in section_activate() */ if (!(pfn & (pageblock_nr_pages - 1))) { @@ -6137,7 +6137,7 @@ void __meminit __weak memmap_init(unsign if (end_pfn > start_pfn) { size = end_pfn - start_pfn; memmap_init_zone(size, nid, zone, start_pfn, - MEMMAP_EARLY, NULL); + MEMINIT_EARLY, NULL); } } }