Skip to contentSkip to conversation

kexec: unmap reserved pages for each error-return way

PATCH In review Dmitry Safonov 5 participants 13 messages Jan 27, 2016 – Feb 2, 2016 2 messages not in the archive

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 kexec: unmap reserved pages for each error-return way

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

---
 arch/s390/kernel/machine_kexec.c | 39 +++++++++++++++++----------------------
 include/linux/kexec.h            |  2 +-
 kernel/kexec.c                   |  4 ++--
 kernel/kexec_core.c              |  4 ++--
 4 files changed, 22 insertions(+), 27 deletions(-)
arch/s390/kernel/machine_kexec.c +17 -22
index 2f1b721..60bf374 100644
@@ -49,7 +49,7 @@ static int machine_kdump_pm_cb(struct notifier_block *nb, unsigned long action,
 	case PM_POST_SUSPEND:
 	case PM_POST_HIBERNATION:
 		if (crashk_res.start)
-			crash_unmap_reserved_pages();
+			crash_unmap_reserved_pages(0);
 		break;
 	default:
 		return NOTIFY_DONE;
@@ -147,39 +147,34 @@ static int kdump_csum_valid(struct kimage *image)
 }
 
 /*
- * Map or unmap crashkernel memory
+ * Map crashkernel memory
  */
-static void crash_map_pages(int enable)
+void crash_map_reserved_pages(void)
 {
 	unsigned long size = resource_size(&crashk_res);
 
 	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
 	       size % KEXEC_CRASH_MEM_ALIGN);
-	if (enable)
-		vmem_add_mapping(crashk_res.start, size);
-	else {
-		vmem_remove_mapping(crashk_res.start, size);
-		if (size)
-			os_info_crashkernel_add(crashk_res.start, size);
-		else
-			os_info_crashkernel_add(0, 0);
-	}
-}
-
-/*
- * Map crashkernel memory
- */
-void crash_map_reserved_pages(void)
-{
-	crash_map_pages(1);
+	vmem_add_mapping(crashk_res.start, size);
 }
 
 /*
  * Unmap crashkernel memory
  */
-void crash_unmap_reserved_pages(void)
+void crash_unmap_reserved_pages(int error)
 {
-	crash_map_pages(0);
+	unsigned long size = resource_size(&crashk_res);
+
+	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
+	       size % KEXEC_CRASH_MEM_ALIGN);
+	vmem_remove_mapping(crashk_res.start, size);
+
+	if (error)
+		return;
+	if (size)
+		os_info_crashkernel_add(crashk_res.start, size);
+	else
+		os_info_crashkernel_add(0, 0);
 }
 
 /*
include/linux/kexec.h +1 -1
index 2cc643c..ef3bd1e 100644
@@ -231,7 +231,7 @@ int kexec_should_crash(struct task_struct *);
 void crash_save_cpu(struct pt_regs *regs, int cpu);
 void crash_save_vmcoreinfo(void);
 void crash_map_reserved_pages(void);
-void crash_unmap_reserved_pages(void);
+void crash_unmap_reserved_pages(int error);
 void arch_crash_save_vmcoreinfo(void);
 __printf(1, 2)
 void vmcoreinfo_append_str(const char *fmt, ...);
kernel/kexec.c +2 -2
index ee70aef..dfc6c18 100644
@@ -204,13 +204,13 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 				goto out;
 		}
 		kimage_terminate(image);
-		if (flags & KEXEC_ON_CRASH)
-			crash_unmap_reserved_pages();
 	}
 	/* Install the new kernel, and  Uninstall the old */
 	image = xchg(dest_image, image);
 
 out:
+	if ((flags & KEXEC_ON_CRASH) && (nr_segments > 0))
+		crash_unmap_reserved_pages(result);
 	mutex_unlock(&kexec_mutex);
 	kimage_free(image);
 
kernel/kexec_core.c +2 -2
index 8dc6591..cae5d11 100644
@@ -965,7 +965,7 @@ int crash_shrink_memory(unsigned long new_size)
 	crashk_res.end = end - 1;
 
 	insert_resource(&iomem_resource, ram_res);
-	crash_unmap_reserved_pages();
+	crash_unmap_reserved_pages(0);
 
 unlock:
 	mutex_unlock(&kexec_mutex);
@@ -1555,5 +1555,5 @@ int kernel_kexec(void)
 void __weak crash_map_reserved_pages(void)
 {}
 
-void __weak crash_unmap_reserved_pages(void)
+void __weak crash_unmap_reserved_pages(int error)
 {}
Signature · 1 line
2.7.0
maintainer
#1 lore raw
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

so can we please sort all that out?

#2 lore raw
Quoted 14 lines · On Wed, 27 Jan 2016 11:15:46 -0800 Andrew Morton <[email protected]>…
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

Quoted 7 lines

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

Hmm, It looks to me that arch_kexec_(un)protect_crashkres() has exactly the same semantics as crash_(un)map_reserved_pages().

On s390 we don't have the crashkernel memory mapped and therefore need crash_map_reserved_pages() before loading something into crashkernel memory.

Perhaps I missed something?
Michael

#3 lore raw
Quoted 21 lines · Michael Holzheu
Quoted 18 lines · On Wed, 27 Jan 2016 11:15:46 -0800 Andrew Morton <[email protected]>…
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

Quoted 8 lines
Quoted 2 lines

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

Hmm, It looks to me that arch_kexec_(un)protect_crashkres() has exactly the same semantics as crash_(un)map_reserved_pages().

On s390 we don't have the crashkernel memory mapped and therefore need crash_map_reserved_pages() before loading something into crashkernel memory.

I don't know s390, just curious, if s390 doesn't have crash kernel memory mapped, what's the purpose of the commit(558df7209e) for s390 as the reserved crash memory with no kernel mapping already means the protection is on?

Regards,
Xunlei

Quoted 9 lines

Perhaps I missed something?
Michael

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

#4 lore raw
Quoted 27 lines · On Thu, 28 Jan 2016 19:56:56 +0800 Xunlei Pang <[email protected]> wrote:
Quoted 26 lines · Michael Holzheu
Quoted 13 lines · On Wed, 27 Jan 2016 11:15:46 -0800 Andrew Morton <[email protected]>…
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

Quoted 7 lines

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

Hmm, It looks to me that arch_kexec_(un)protect_crashkres() has exactly the same semantics as crash_(un)map_reserved_pages().

Quoted 7 lines
Quoted 3 lines

On s390 we don't have the crashkernel memory mapped and therefore need crash_map_reserved_pages() before loading something into crashkernel memory.

I don't know s390, just curious, if s390 doesn't have crash kernel memory mapped, what's the purpose of the commit(558df7209e) for s390 as the reserved crash memory with no kernel mapping already means the protection is on?

When we reserve crashkernel memory on s390 ("crashkernel=" kernel parameter), we create a memory hole without page tables.

Commit (558df7209e) was necessary to load a kernel/ramdisk into the memory hole with the kexec() system call.

We create a temporary mapping with crash_map_reserved_pages(), then copy the kernel/ramdisk and finally remove the mapping again via crash_unmap_reserved_pages().

We did that all in order to protect the preloaded kernel and ramdisk.

I forgot the details why commit(558df7209e) wasn't necessary before. AFAIK it became necessary because of some kdump (mmap?) rework.

Michael

#5 lore raw
Quoted 39 lines · Michael Holzheu
Quoted 27 lines · On Thu, 28 Jan 2016 19:56:56 +0800 Xunlei Pang <[email protected]> wrote:
Quoted 26 lines · Michael Holzheu
Quoted 13 lines · On Wed, 27 Jan 2016 11:15:46 -0800 Andrew Morton <[email protected]>…
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

Quoted 7 lines

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

Hmm, It looks to me that arch_kexec_(un)protect_crashkres() has exactly the same semantics as crash_(un)map_reserved_pages().

Quoted 6 lines
Quoted 3 lines

On s390 we don't have the crashkernel memory mapped and therefore need crash_map_reserved_pages() before loading something into crashkernel memory.

I don't know s390, just curious, if s390 doesn't have crash kernel memory mapped, what's the purpose of the commit(558df7209e) for s390 as the reserved crash memory with no kernel mapping already means the protection is on?

When we reserve crashkernel memory on s390 ("crashkernel=" kernel parameter), we create a memory hole without page tables.

Quoted 6 lines

Commit (558df7209e) was necessary to load a kernel/ramdisk into the memory hole with the kexec() system call.

We create a temporary mapping with crash_map_reserved_pages(), then copy the kernel/ramdisk and finally remove the mapping again via crash_unmap_reserved_pages().

Thanks for the explanation.
So, on s390 the physical memory address has the same value as its kernel virtual address, and kmap() actually returns the value of the physical address of the page, right?

Quoted 5 lines

We did that all in order to protect the preloaded kernel and ramdisk.

I forgot the details why commit(558df7209e) wasn't necessary before. AFAIK it became necessary because of some kdump (mmap?) rework.

Uh, this is indeed strange.

Regards,
Xunlei

Quoted 3 lines

Michael

#6 lore raw
Quoted 43 lines · On Thu, 28 Jan 2016 21:12:54 +0800 Xunlei Pang <[email protected]> wrote:
Quoted 42 lines · Michael Holzheu
Quoted 27 lines · On Thu, 28 Jan 2016 19:56:56 +0800 Xunlei Pang <[email protected]> wrote:
Quoted 26 lines · Michael Holzheu
Quoted 13 lines · On Wed, 27 Jan 2016 11:15:46 -0800 Andrew Morton <[email protected]>…
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

Quoted 7 lines

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

Hmm, It looks to me that arch_kexec_(un)protect_crashkres() has exactly the same semantics as crash_(un)map_reserved_pages().

Quoted 6 lines
Quoted 3 lines

On s390 we don't have the crashkernel memory mapped and therefore need crash_map_reserved_pages() before loading something into crashkernel memory.

I don't know s390, just curious, if s390 doesn't have crash kernel memory mapped, what's the purpose of the commit(558df7209e) for s390 as the reserved crash memory with no kernel mapping already means the protection is on?

When we reserve crashkernel memory on s390 ("crashkernel=" kernel parameter), we create a memory hole without page tables.

Commit (558df7209e) was necessary to load a kernel/ramdisk into the memory hole with the kexec() system call.

Quoted 7 lines
Quoted 3 lines

We create a temporary mapping with crash_map_reserved_pages(), then copy the kernel/ramdisk and finally remove the mapping again via crash_unmap_reserved_pages().

Thanks for the explanation.
So, on s390 the physical memory address has the same value as its kernel virtual address, and kmap() actually returns the value of the physical address of the page, right?

Correct. On s390 kmap() always return the physical address of the page.

We have an 1:1 mapping for all the physical memory. For this area virtual=real. In addition to that we have the vmalloc area above the 1:1 mapping where some of the memory is mapped a second time.

Michael

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#9 lore raw
Quoted 25 lines · Dmitry Safonov
Quoted 24 lines · Xunlei Pang

Hi Dmitry,

Quoted 21 lines · Andrew Morton
Quoted 6 lines · On Wed, 27 Jan 2016 14:48:31 +0300 Dmitry Safonov <[email protected]>…

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

This conflicts (both mechanically and somewhat conceptually) with Xunlei Pang's "kexec: Introduce a protection mechanism for the crashkernel reserved memory" and "kexec: provide arch_kexec_protect(unprotect)_crashkres()".

http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-introduce-a-protection-mechanism-for-the-crashkernel-reserved-memory-v4.patch

and

http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres.patch
http://ozlabs.org/~akpm/mmots/broken-out/kexec-provide-arch_kexec_protectunprotect_crashkres-v4.patch

Quoted 6 lines
Quoted 4 lines
Quoted 1 line

so can we please sort all that out?

Ah, I've checked git-log history, they are almost the same idea, can crash_unmap/map_reserved_pages() be re-implemented using the new arch_kexec_unprotect/protect_crashkres() on S390?

Sorry, didn't fetched akpm before sending.
Yes, sounds like really right thing to do to have one united arch-helper.

Yeah, as Michael said, "memblock_remove(crash_base, crash_size)" creates a big hole in the kernel pgtable. In order to have one united arch-helper, I guess we can forbid this to let the pgtable setup for crash memory, then we can easily move the logic of crash_map_reserved_pages() to arch_kexec_unprotect_crashkres(), and move crash_unmap_reserved_pages() to arch_kexec_protect_crashkres(). After that crash_map_reserved_pages() called in crash_shrink_memory() can be safely removed as well.

Regards,
Xunlei

Quoted 14 lines
Quoted 9 lines

Regards,
Xunlei

Quoted 4 lines

_______________________________________________
kexec mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/kexec

Signature · 2 lines
Regards,
Dmitry Safonov
#10 lore raw
In reply to Dmitry Safonov
Quoted 71 lines · Dmitry Safonov

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

---
 arch/s390/kernel/machine_kexec.c | 39 +++++++++++++++++----------------------
 include/linux/kexec.h            |  2 +-
 kernel/kexec.c                   |  4 ++--
 kernel/kexec_core.c              |  4 ++--
 4 files changed, 22 insertions(+), 27 deletions(-)
arch/s390/kernel/machine_kexec.c +8 -22
index 2f1b721..60bf374 100644
@@ -49,7 +49,7 @@ static int machine_kdump_pm_cb(struct notifier_block *nb, unsigned long action,
 	case PM_POST_SUSPEND:
 	case PM_POST_HIBERNATION:
 		if (crashk_res.start)
-			crash_unmap_reserved_pages();
+			crash_unmap_reserved_pages(0);
 		break;
 	default:
 		return NOTIFY_DONE;
@@ -147,39 +147,34 @@ static int kdump_csum_valid(struct kimage *image)
 }
 
 /*
- * Map or unmap crashkernel memory
+ * Map crashkernel memory
  */
-static void crash_map_pages(int enable)
+void crash_map_reserved_pages(void)
 {
 	unsigned long size = resource_size(&crashk_res);
 
 	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
 	       size % KEXEC_CRASH_MEM_ALIGN);
-	if (enable)
-		vmem_add_mapping(crashk_res.start, size);
-	else {
-		vmem_remove_mapping(crashk_res.start, size);
-		if (size)
-			os_info_crashkernel_add(crashk_res.start, size);
-		else
-			os_info_crashkernel_add(0, 0);
-	}
-}
-
-/*
- * Map crashkernel memory
- */
-void crash_map_reserved_pages(void)
-{
-	crash_map_pages(1);
+	vmem_add_mapping(crashk_res.start, size);
 }
 
 /*
  * Unmap crashkernel memory
  */
-void crash_unmap_reserved_pages(void)
+void crash_unmap_reserved_pages(int error)
 {
-	crash_map_pages(0);
+	unsigned long size = resource_size(&crashk_res);
+
+	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||

arch/s390/kernel/machine_kexec.c static int kdump_csum_valid(struct kimage *image) · line 178

Quoted 10 lines
arch/s390/kernel/machine_kexec.c +9 -0
+	       size % KEXEC_CRASH_MEM_ALIGN);
+	vmem_remove_mapping(crashk_res.start, size);
+
+	if (error)
+		return;
+	if (size)
+		os_info_crashkernel_add(crashk_res.start, size);
+	else
+		os_info_crashkernel_add(0, 0);
 }

I think it doesn't hurt doing os_info_crashkernel_add() multiple times which will make the code cleaner.

Quoted 24 lines
arch/s390/kernel/machine_kexec.c +0 -0
 /*
include/linux/kexec.h +1 -1
index 2cc643c..ef3bd1e 100644
@@ -231,7 +231,7 @@ int kexec_should_crash(struct task_struct *);
 void crash_save_cpu(struct pt_regs *regs, int cpu);
 void crash_save_vmcoreinfo(void);
 void crash_map_reserved_pages(void);
-void crash_unmap_reserved_pages(void);
+void crash_unmap_reserved_pages(int error);
 void arch_crash_save_vmcoreinfo(void);
 __printf(1, 2)
 void vmcoreinfo_append_str(const char *fmt, ...);
kernel/kexec.c +0 -1
index ee70aef..dfc6c18 100644
@@ -204,13 +204,13 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments,
 				goto out;
 		}
 		kimage_terminate(image);
-		if (flags & KEXEC_ON_CRASH)

kernel/kexec.c SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments, · line 215

Quoted 10 lines
kernel/kexec.c +2 -1
-			crash_unmap_reserved_pages();
 	}
 	/* Install the new kernel, and  Uninstall the old */
 	image = xchg(dest_image, image);
 
 out:
+	if ((flags & KEXEC_ON_CRASH) && (nr_segments > 0))
+		crash_unmap_reserved_pages(result);
 	mutex_unlock(&kexec_mutex);
 	kimage_free(image);

Shouldn't we do the similar thing for kexec_file_load()?

kernel/kexec_core.c int crash_shrink_memory(unsigned long new_size) · line 968

Quoted 11 lines
kernel/kexec_core.c +1 -1
index 8dc6591..cae5d11 100644
@@ -965,7 +965,7 @@ int crash_shrink_memory(unsigned long new_size)
 	crashk_res.end = end - 1;
 
 	insert_resource(&iomem_resource, ram_res);
-	crash_unmap_reserved_pages();
+	crash_unmap_reserved_pages(0);

If using arch_kexec_unprotect/protect_crashkres(), we don't need this since there's a judge at the entry :-)

    if (kexec_crash_image) {
        ret = -ENOENT;
        goto unlock;
    }

Regards,
Xunlei

Quoted 10 lines
kernel/kexec_core.c +1 -1
 unlock:
 	mutex_unlock(&kexec_mutex);
@@ -1555,5 +1555,5 @@ int kernel_kexec(void)
 void __weak crash_map_reserved_pages(void)
 {}
 
-void __weak crash_unmap_reserved_pages(void)
+void __weak crash_unmap_reserved_pages(int error)
 {}
#11 lore raw
In reply to Dmitry Safonov
Quoted 5 lines · Dmitry Safonov

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

Hi, Dmitry.

Previously, I sent a patch to fix this issue. You can refer it in following link.

http://lists.infradead.org/pipermail/kexec/2015-July/013960.html

And this patch is fixed from kexec.

If crash_map_reserved_pages fails to map reserved memory, is it necessary to continue the process on s390? If no, it is better to enter the error handle path, then return. Thus there is no need to pass the parameter to indicate the error or not.

Quoted 23 lines
arch/s390/kernel/machine_kexec.c +2 -11
@@ -147,39 +147,34 @@ static int kdump_csum_valid(struct kimage *image)
 }
 
 /*
- * Map or unmap crashkernel memory
+ * Map crashkernel memory
  */
-static void crash_map_pages(int enable)
+void crash_map_reserved_pages(void)
 {
 	unsigned long size = resource_size(&crashk_res);
 
 	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
 	       size % KEXEC_CRASH_MEM_ALIGN);
-	if (enable)
-		vmem_add_mapping(crashk_res.start, size);
-	else {
-		vmem_remove_mapping(crashk_res.start, size);
-		if (size)
-			os_info_crashkernel_add(crashk_res.start, size);
-		else
-			os_info_crashkernel_add(0, 0);
-	}

arch/s390/kernel/machine_kexec.c static int kdump_csum_valid(struct kimage *image) · line 159

Quoted 10 lines
arch/s390/kernel/machine_kexec.c +1 -8
-}
-
-/*
- * Map crashkernel memory
- */
-void crash_map_reserved_pages(void)
-{
-	crash_map_pages(1);
+	vmem_add_mapping(crashk_res.start, size);
 }

It is fine to cleanup this function. And add the logic into function crash_unmap_reserved_pages.

Quoted 13 lines
arch/s390/kernel/machine_kexec.c +6 -2
 /*
  * Unmap crashkernel memory
  */
-void crash_unmap_reserved_pages(void)
+void crash_unmap_reserved_pages(int error)
 {
-	crash_map_pages(0);
+	unsigned long size = resource_size(&crashk_res);
+
+	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
+	       size % KEXEC_CRASH_MEM_ALIGN);
+	vmem_remove_mapping(crashk_res.start, size);

arch/s390/kernel/machine_kexec.c static int kdump_csum_valid(struct kimage *image) · line 180

Quoted 10 lines
arch/s390/kernel/machine_kexec.c +7 -0
+
+	if (error)
+		return;
+	if (size)
+		os_info_crashkernel_add(crashk_res.start, size);
+	else
+		os_info_crashkernel_add(0, 0);
 }
 
 /*

Thanks
Minfei

#12 lore raw
Quoted 6 lines · Minfei Huang
Quoted 5 lines · Dmitry Safonov

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

Quoted 6 lines

Hi, Dmitry.

Previously, I sent a patch to fix this issue. You can refer it in following link.

http://lists.infradead.org/pipermail/kexec/2015-July/013960.html

Oh, scratch my patch - I'm fine with yours, wanted to do the similar thing because it has dazzled me while I was debugging around.

Quoted 69 lines

And this patch is fixed from kexec.

If crash_map_reserved_pages fails to map reserved memory, is it necessary to continue the process on s390? If no, it is better to enter the error handle path, then return. Thus there is no need to pass the parameter to indicate the error or not.

Quoted 23 lines
arch/s390/kernel/machine_kexec.c +2 -11
@@ -147,39 +147,34 @@ static int kdump_csum_valid(struct kimage *image)
  }
 
  /*
- * Map or unmap crashkernel memory
+ * Map crashkernel memory
   */
-static void crash_map_pages(int enable)
+void crash_map_reserved_pages(void)
  {
  	unsigned long size = resource_size(&crashk_res);
 
  	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
  	       size % KEXEC_CRASH_MEM_ALIGN);
-	if (enable)
-		vmem_add_mapping(crashk_res.start, size);
-	else {
-		vmem_remove_mapping(crashk_res.start, size);
-		if (size)
-			os_info_crashkernel_add(crashk_res.start, size);
-		else
-			os_info_crashkernel_add(0, 0);
-	}

arch/s390/kernel/machine_kexec.c static int kdump_csum_valid(struct kimage *image) · line 159

Quoted 10 lines
arch/s390/kernel/machine_kexec.c +1 -8
-}
-
-/*
- * Map crashkernel memory
- */
-void crash_map_reserved_pages(void)
-{
-	crash_map_pages(1);
+	vmem_add_mapping(crashk_res.start, size);
  }

It is fine to cleanup this function. And add the logic into function crash_unmap_reserved_pages.

Quoted 13 lines
arch/s390/kernel/machine_kexec.c +6 -2
  /*
   * Unmap crashkernel memory
   */
-void crash_unmap_reserved_pages(void)
+void crash_unmap_reserved_pages(int error)
  {
-	crash_map_pages(0);
+	unsigned long size = resource_size(&crashk_res);
+
+	BUG_ON(crashk_res.start % KEXEC_CRASH_MEM_ALIGN ||
+	       size % KEXEC_CRASH_MEM_ALIGN);
+	vmem_remove_mapping(crashk_res.start, size);

arch/s390/kernel/machine_kexec.c static int kdump_csum_valid(struct kimage *image) · line 180

Quoted 10 lines
arch/s390/kernel/machine_kexec.c +7 -0
+
+	if (error)
+		return;
+	if (size)
+		os_info_crashkernel_add(crashk_res.start, size);
+	else
+		os_info_crashkernel_add(0, 0);
  }
 
  /*

Thanks
Minfei

Signature · 2 lines
Regards,
Dmitry Safonov
maintainer
#13 lore raw
Quoted 7 lines · On Thu, 28 Jan 2016 11:57:22 +0300 Dmitry Safonov <[email protected]>…
Quoted 6 lines · Minfei Huang
Quoted 5 lines · Dmitry Safonov

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

Quoted 8 lines
Quoted 6 lines

Hi, Dmitry.

Previously, I sent a patch to fix this issue. You can refer it in following link.

http://lists.infradead.org/pipermail/kexec/2015-July/013960.html

Oh, scratch my patch - I'm fine with yours, wanted to do the similar thing because it has dazzled me while I was debugging around.

There were a bunch of patches tossed around in that thread but I'm not sure that anything actually got applied? Perhaps some resending is needed.

#14 lore raw
Quoted 14 lines · Andrew Morton
Quoted 12 lines · On Thu, 28 Jan 2016 11:57:22 +0300 Dmitry Safonov <[email protected]>…
Quoted 11 lines · Minfei Huang
Quoted 5 lines · Dmitry Safonov

For allocation of kimage failure or kexec_prepare or load segments errors there is no need to keep crashkernel memory mapped. It will affect only s390 as map/unmap hook defined only for it. As on unmap s390 also changes os_info structure let's check return code and add info only on success.

Hi, Dmitry.

Previously, I sent a patch to fix this issue. You can refer it in following link.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

http://lists.infradead.org/pipermail/kexec/2015-July/013960.html

Oh, scratch my patch - I'm fine with yours, wanted to do the similar thing because it has dazzled me while I was debugging around.

There were a bunch of patches tossed around in that thread but I'm not sure that anything actually got applied? Perhaps some resending is needed.

Hi, Andrew.

I will work on it to update a new patch.

Thanks
Minfei