On Thu, 2019-08-15 at 09:36 +0200, christophe leroy wrote:

Le 15/08/2019 à 06:10, Alastair D'Silva a écrit :

quoted

From: Alastair D'Silva <redacted>

When presented with large amounts of memory being hotplugged
(in my test case, ~890GB), the call to flush_dcache_range takes
a while (~50 seconds), triggering RCU stalls.

This patch breaks up the call into 16GB chunks, calling
cond_resched() inbetween to allow the scheduler to run.

Is 16GB small enough ? If 890GB takes 50s, 16GB still takes about 1s.
I'd use 1GB chuncks to remain below 100ms.

quoted

Signed-off-by: Alastair D'Silva <redacted>
---
  arch/powerpc/mm/mem.c | 16 ++++++++++++++--
  1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index 5400da87a804..fb0d5e9aa11b 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c

@@ -104,11 +104,14 @@ int __weak remove_section_mapping(unsigned

long start, unsigned long end)
  	return -ENODEV;
  }
  
+#define FLUSH_CHUNK_SIZE (16ull * 1024ull * 1024ull * 1024ull)

Can we use SZ_16GB ?

Sure, I'll go with 1GB as you recommended above

quoted

+
  int __ref arch_add_memory(int nid, u64 start, u64 size,
  			struct mhp_restrictions *restrictions)
  {
  	unsigned long start_pfn = start >> PAGE_SHIFT;
  	unsigned long nr_pages = size >> PAGE_SHIFT;
+	unsigned long i;
  	int rc;
  
  	resize_hpt_for_hotplug(memblock_phys_mem_size());

@@ -120,7 +123,11 @@ int __ref arch_add_memory(int nid, u64 start,

u64 size,
  			start, start + size, rc);
  		return -EFAULT;
  	}
-	flush_dcache_range(start, start + size);
+
+	for (i = 0; i < size; i += FLUSH_CHUNK_SIZE) {
+		flush_dcache_range(start + i, min(start + size, start +
i + FLUSH_CHUNK_SIZE));

Isn't the line a bit long (I have not checked).

quoted

+		cond_resched();
+	}
  
  	return __add_pages(nid, start_pfn, nr_pages, restrictions);
  }

@@ -131,13 +138,18 @@ void __ref arch_remove_memory(int nid, u64

start, u64 size,
  	unsigned long start_pfn = start >> PAGE_SHIFT;
  	unsigned long nr_pages = size >> PAGE_SHIFT;
  	struct page *page = pfn_to_page(start_pfn) +
vmem_altmap_offset(altmap);
+	unsigned long i;
  	int ret;
  
  	__remove_pages(page_zone(page), start_pfn, nr_pages, altmap);
  
  	/* Remove htab bolted mappings for this section of memory */
  	start = (unsigned long)__va(start);
-	flush_dcache_range(start, start + size);
+	for (i = 0; i < size; i += FLUSH_CHUNK_SIZE) {
+		flush_dcache_range(start + i, min(start + size, start +
i + FLUSH_CHUNK_SIZE));
+		cond_resched();
+	}
+
  	ret = remove_section_mapping(start, start + size);
  	WARN_ON_ONCE(ret);
  

Christophe

---
L'absence de virus dans ce courrier électronique a été vérifiée par
le logiciel antivirus Avast.
https://urldefense.proofpoint.com/v2/url?u=https-3A__www.avast.com_antivirus&d=DwIDaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=cT4tgeEQ0Ll3SIlZDHE5AEXyKy6uKADMtf9_Eb7-vec&m=TBT2NNM2DXqDWHhSb_WdFPcfAjYk9hP2cvGksF001cQ&s=XURKAOQQ4h3_RhJlezSguD2kpSitAF-uBhQqVZLU4GU&e= 

-- 
Alastair D'Silva
Open Source Developer
Linux Technology Centre, IBM Australia
mob: 0423 762 819