A memory leak is reported by kmemleak scanning:
unreferenced object 0x38000195000 (size 4096):
comm "kexec", pid 8548, jiffies 4294953647 (age 32443.270s)
hex dump (first 32 bytes):
00 00 00 c8 20 00 00 00 00 00 00 c0 02 80 00 00 .... ...........
40 40 40 40 40 40 40 40 00 00 00 00 00 00 00 00 @@@@@@@@........
backtrace:
[<0000000011a2f199>] __vmalloc_node_range+0xc0/0x140
[<0000000081fa2752>] vzalloc+0x5a/0x70
[<0000000063a4c92d>] ipl_report_finish+0x2c/0x180
[<00000000553304da>] kexec_file_add_ipl_report+0xf4/0x150
[<00000000862d033f>] kexec_file_add_components+0x124/0x160
[<000000000d2717bb>] arch_kexec_kernel_image_load+0x62/0x90
[<000000002e0373b6>] kimage_file_alloc_init+0x1aa/0x2e0
[<0000000060f2d14f>] __do_sys_kexec_file_load+0x17c/0x2c0
[<000000008c86fe5a>] __s390x_sys_kexec_file_load+0x40/0x50
[<000000001fdb9dac>] __do_syscall+0x1bc/0x1f0
[<000000003ee4258d>] system_call+0x78/0xa0
The ipl report buffer is allocated via vmalloc, while has no chance to free
if the kexe loading is unloaded. This will cause obvious memory leak
when kexec/kdump kernel is reloaded, manually, or triggered, e.g by
memory hotplug.
Here, add struct kimage_arch to s390 to pass out the ipl report buffer
address, and introduce arch dependent function
arch_kimage_file_post_load_cleanup() to free it when unloaded.
Signed-off-by: Baoquan He <[email protected]>
---
arch/s390/include/asm/kexec.h | 7 +++++++
arch/s390/kernel/machine_kexec_file.c | 9 +++++++++
2 files changed, 16 insertions(+)
diff --git a/arch/s390/include/asm/kexec.h b/arch/s390/include/asm/kexec.h
index ea398a05f643..9f16e99fb882 100644
--- a/arch/s390/include/asm/kexec.h
+++ b/arch/s390/include/asm/kexec.h
@@ -74,6 +74,13 @@ void *kexec_file_add_components(struct kimage *image,
int arch_kexec_do_relocs(int r_type, void *loc, unsigned long val,
unsigned long addr);
+#define ARCH_HAS_KIMAGE_ARCH
+
+struct kimage_arch {
+ void *ipl_buf;
+};
+
+
extern const struct kexec_file_ops s390_kexec_image_ops;
extern const struct kexec_file_ops s390_kexec_elf_ops;
diff --git a/arch/s390/kernel/machine_kexec_file.c b/arch/s390/kernel/machine_kexec_file.c
index f9e4baa64b67..584c48b631b2 100644
--- a/arch/s390/kernel/machine_kexec_file.c
+++ b/arch/s390/kernel/machine_kexec_file.c
@@ -202,6 +202,7 @@ static int kexec_file_add_ipl_report(struct kimage *image,
buf.buffer = ipl_report_finish(data->report);
buf.bufsz = data->report->size;
buf.memsz = buf.bufsz;
+ image->arch.ipl_buf = buf.buffer;
data->memsz += buf.memsz;
@@ -321,3 +322,11 @@ int arch_kexec_kernel_image_probe(struct kimage *image, void *buf,
return kexec_image_probe_default(image, buf, buf_len);
}
+
+int arch_kimage_file_post_load_cleanup(struct kimage *image)
+{
+ kvfree(image->arch.ipl_buf);
+ image->arch.ipl_buf = NULL;
+
+ return kexec_image_post_load_cleanup_default(image);
+}
--
2.17.2
Hi Baoquan,
On Fri, 29 Oct 2021 17:26:35 +0800
Baoquan He <[email protected]> wrote:
> A memory leak is reported by kmemleak scanning:
>
> unreferenced object 0x38000195000 (size 4096):
> comm "kexec", pid 8548, jiffies 4294953647 (age 32443.270s)
> hex dump (first 32 bytes):
> 00 00 00 c8 20 00 00 00 00 00 00 c0 02 80 00 00 .... ...........
> 40 40 40 40 40 40 40 40 00 00 00 00 00 00 00 00 @@@@@@@@........
> backtrace:
> [<0000000011a2f199>] __vmalloc_node_range+0xc0/0x140
> [<0000000081fa2752>] vzalloc+0x5a/0x70
> [<0000000063a4c92d>] ipl_report_finish+0x2c/0x180
> [<00000000553304da>] kexec_file_add_ipl_report+0xf4/0x150
> [<00000000862d033f>] kexec_file_add_components+0x124/0x160
> [<000000000d2717bb>] arch_kexec_kernel_image_load+0x62/0x90
> [<000000002e0373b6>] kimage_file_alloc_init+0x1aa/0x2e0
> [<0000000060f2d14f>] __do_sys_kexec_file_load+0x17c/0x2c0
> [<000000008c86fe5a>] __s390x_sys_kexec_file_load+0x40/0x50
> [<000000001fdb9dac>] __do_syscall+0x1bc/0x1f0
> [<000000003ee4258d>] system_call+0x78/0xa0
>
> The ipl report buffer is allocated via vmalloc, while has no chance to free
> if the kexe loading is unloaded. This will cause obvious memory leak
> when kexec/kdump kernel is reloaded, manually, or triggered, e.g by
> memory hotplug.
>
> Here, add struct kimage_arch to s390 to pass out the ipl report buffer
> address, and introduce arch dependent function
> arch_kimage_file_post_load_cleanup() to free it when unloaded.
>
> Signed-off-by: Baoquan He <[email protected]>
other than a missing
Fixes: 99feaa717e55 ("s390/kexec_file: Create ipl report and pass to
next kernel")
the patch looks good to me.
Reviewed-by: Philipp Rudo <[email protected]>
> ---
> arch/s390/include/asm/kexec.h | 7 +++++++
> arch/s390/kernel/machine_kexec_file.c | 9 +++++++++
> 2 files changed, 16 insertions(+)
>
> diff --git a/arch/s390/include/asm/kexec.h b/arch/s390/include/asm/kexec.h
> index ea398a05f643..9f16e99fb882 100644
> --- a/arch/s390/include/asm/kexec.h
> +++ b/arch/s390/include/asm/kexec.h
> @@ -74,6 +74,13 @@ void *kexec_file_add_components(struct kimage *image,
> int arch_kexec_do_relocs(int r_type, void *loc, unsigned long val,
> unsigned long addr);
>
> +#define ARCH_HAS_KIMAGE_ARCH
> +
> +struct kimage_arch {
> + void *ipl_buf;
> +};
> +
> +
> extern const struct kexec_file_ops s390_kexec_image_ops;
> extern const struct kexec_file_ops s390_kexec_elf_ops;
>
> diff --git a/arch/s390/kernel/machine_kexec_file.c b/arch/s390/kernel/machine_kexec_file.c
> index f9e4baa64b67..584c48b631b2 100644
> --- a/arch/s390/kernel/machine_kexec_file.c
> +++ b/arch/s390/kernel/machine_kexec_file.c
> @@ -202,6 +202,7 @@ static int kexec_file_add_ipl_report(struct kimage *image,
> buf.buffer = ipl_report_finish(data->report);
> buf.bufsz = data->report->size;
> buf.memsz = buf.bufsz;
> + image->arch.ipl_buf = buf.buffer;
>
> data->memsz += buf.memsz;
>
> @@ -321,3 +322,11 @@ int arch_kexec_kernel_image_probe(struct kimage *image, void *buf,
>
> return kexec_image_probe_default(image, buf, buf_len);
> }
> +
> +int arch_kimage_file_post_load_cleanup(struct kimage *image)
> +{
> + kvfree(image->arch.ipl_buf);
> + image->arch.ipl_buf = NULL;
> +
> + return kexec_image_post_load_cleanup_default(image);
> +}
On Fri, Oct 29, 2021 at 06:31:32PM +0200, Philipp Rudo wrote:
> Hi Baoquan,
>
> On Fri, 29 Oct 2021 17:26:35 +0800
> Baoquan He <[email protected]> wrote:
>
> > A memory leak is reported by kmemleak scanning:
...
> > The ipl report buffer is allocated via vmalloc, while has no chance to free
> > if the kexe loading is unloaded. This will cause obvious memory leak
> > when kexec/kdump kernel is reloaded, manually, or triggered, e.g by
> > memory hotplug.
> >
> > Here, add struct kimage_arch to s390 to pass out the ipl report buffer
> > address, and introduce arch dependent function
> > arch_kimage_file_post_load_cleanup() to free it when unloaded.
> >
> > Signed-off-by: Baoquan He <[email protected]>
>
> other than a missing
>
> Fixes: 99feaa717e55 ("s390/kexec_file: Create ipl report and pass to
> next kernel")
>
> the patch looks good to me.
>
> Reviewed-by: Philipp Rudo <[email protected]>
...
> > buf.buffer = ipl_report_finish(data->report);
> > buf.bufsz = data->report->size;
> > buf.memsz = buf.bufsz;
> > + image->arch.ipl_buf = buf.buffer;
This seems (still) to be incorrect: ipl_report_finish() may return
-ENOMEN, but there is no error checking anywhere, as far as I can
tell, which would make this:
> > +int arch_kimage_file_post_load_cleanup(struct kimage *image)
> > +{
> > + kvfree(image->arch.ipl_buf);
> > + image->arch.ipl_buf = NULL;
> > +
> > + return kexec_image_post_load_cleanup_default(image);
> > +}
most likely not do what we want. That is: if this code is reached at
all in such a case. I'll check and might add a patch before this to
fix this also.
On 11/10/21 at 04:18pm, Heiko Carstens wrote:
> On Fri, Oct 29, 2021 at 06:31:32PM +0200, Philipp Rudo wrote:
> > Hi Baoquan,
> >
> > On Fri, 29 Oct 2021 17:26:35 +0800
> > Baoquan He <[email protected]> wrote:
> >
> > > A memory leak is reported by kmemleak scanning:
> ...
> > > The ipl report buffer is allocated via vmalloc, while has no chance to free
> > > if the kexe loading is unloaded. This will cause obvious memory leak
> > > when kexec/kdump kernel is reloaded, manually, or triggered, e.g by
> > > memory hotplug.
> > >
> > > Here, add struct kimage_arch to s390 to pass out the ipl report buffer
> > > address, and introduce arch dependent function
> > > arch_kimage_file_post_load_cleanup() to free it when unloaded.
> > >
> > > Signed-off-by: Baoquan He <[email protected]>
> >
> > other than a missing
> >
> > Fixes: 99feaa717e55 ("s390/kexec_file: Create ipl report and pass to
> > next kernel")
> >
> > the patch looks good to me.
> >
> > Reviewed-by: Philipp Rudo <[email protected]>
> ...
> > > buf.buffer = ipl_report_finish(data->report);
> > > buf.bufsz = data->report->size;
> > > buf.memsz = buf.bufsz;
> > > + image->arch.ipl_buf = buf.buffer;
>
> This seems (still) to be incorrect: ipl_report_finish() may return
> -ENOMEN, but there is no error checking anywhere, as far as I can
> tell, which would make this:
>
> > > +int arch_kimage_file_post_load_cleanup(struct kimage *image)
> > > +{
> > > + kvfree(image->arch.ipl_buf);
> > > + image->arch.ipl_buf = NULL;
> > > +
> > > + return kexec_image_post_load_cleanup_default(image);
> > > +}
>
> most likely not do what we want. That is: if this code is reached at
> all in such a case. I'll check and might add a patch before this to
> fix this also.
Right, we should check the returned value firstly. Thanks a lot for
reivewing, Heiko. I will post v2 to add a patch to check the returned
value as you suggested, and also update this patch to add missing Fixes tag.