[MAKDUMPFILE PATCH] Add option to estimate the size of vmcore dump files
Julien Thierry
jthierry at redhat.com
Tue Oct 20 07:36:41 EDT 2020
Hi Kazuhito,
On 10/16/20 7:45 AM, HAGIO KAZUHITO(萩尾 一仁) wrote:
> Hi Julien,
>
> -----Original Message-----
>> Hello Julien,
>>
>> On Tue, Oct 13, 2020 at 3:23 PM Julien Thierry <jthierry at redhat.com> wrote:
>>>
>>> Hi Bhupesh,
>>>
>>> On 10/13/20 10:27 AM, Bhupesh Sharma wrote:
>>>> Hello Julien,
>>>>
>>>> Thanks for the patch. Some nitpicks inline:
>>>>
>>>> On Mon, Oct 12, 2020 at 12:39 PM Julien Thierry <jthierry at redhat.com> wrote:
>>>>>
>>>>> A user might want to know how much space a vmcore file will take on
>>>>> the system and how much space on their disk should be available to
>>>>> save it during a crash.
>>>>>
>>>>> The option --vmcore-size does not create the vmcore file but provides
>>>>> an estimation of the size of the final vmcore file created with the
>>>>> same make dumpfile options.
>
> Interesting. Do you have any actual use case? e.g. used by kdumpctl?
> or use it in kdump initramfs?
>
Yes, the idea would be to use this in mkdumprd to have a more accurate
estimate of the dump size (currently it cannot take compression into
account and warns about potential lack of space, considering the system
memory size as a whole).
>>>>>
>>>>> Signed-off-by: Julien Thierry <jthierry at redhat.com>
>>>>> Cc: Kazuhito Hagio <k-hagio-ab at nec.com>
>>>>> ---
>>>>> makedumpfile.c | 98 ++++++++++++++++++++++++++++++++++++++++++++++++--
>>>>> makedumpfile.h | 12 +++++++
>>>>> print_info.c | 4 +++
>>>>> 3 files changed, 111 insertions(+), 3 deletions(-)
>>>>
>>>> Please update 'makedumpfile.8' as well in v2, so that the man page can
>>>> document the newly added option and how to use it to determine the
>>>> vmcore-size.
>>>>
>>>
>>> Ah yes, I'll do that.
>>>
>>>>> diff --git a/makedumpfile.c b/makedumpfile.c
>>>>> index 4c4251e..0a2bfba 100644
>>>>> --- a/makedumpfile.c
>>>>> +++ b/makedumpfile.c
>>>>> @@ -26,6 +26,7 @@
>>>>> #include <limits.h>
>>>>> #include <assert.h>
>>>>> #include <zlib.h>
>>>>> +#include <libgen.h>
>>>>
>>>> I know we don't follow alphabetical order for include files in
>>>> makedumpfile code, but it would be good to place the new - ones
>>>> accordingly. So <libgen.h> can go with <limits.h> here.
>>>>
>>>
>>> Noted.
>>>
>>>>> struct symbol_table symbol_table;
>>>>> struct size_table size_table;
>>>>> @@ -1366,7 +1367,25 @@ open_dump_file(void)
>>>>> if (!info->flag_force)
>>>>> open_flags |= O_EXCL;
>>>>>
>>>>> - if (info->flag_flatten) {
>>>>> + if (info->flag_vmcore_size) {
>>>>> + char *namecpy;
>>>>> + struct stat statbuf;
>>>>> + int res;
>>>>> +
>>>>> + namecpy = strdup(info->name_dumpfile ?
>>>>> + info->name_dumpfile : ".");
>>>>> +
>>>>> + res = stat(dirname(namecpy), &statbuf);
>>>>> + free(namecpy);
>>>>> + if (res != 0)
>>>>> + return FALSE;
>>>>> +
>>>>> + fd = -1;
>>>>> + info->dumpsize_info.blksize = statbuf.st_blksize;
>>>>> + info->dumpsize_info.block_buff_size = BASE_NUM_BLOCKS;
>>>>> + info->dumpsize_info.block_info = calloc(BASE_NUM_BLOCKS, 1);
>>>>> + info->dumpsize_info.non_hole_blocks = 0;
>>>>> + } else if (info->flag_flatten) {
>>>>> fd = STDOUT_FILENO;
>>>>> info->name_dumpfile = filename_stdout;
>>>>> } else if ((fd = open(info->name_dumpfile, open_flags,
>>>>> @@ -1384,6 +1403,9 @@ check_dump_file(const char *path)
>>>>> {
>>>>> char *err_str;
>>>>>
>>>>> + if (info->flag_vmcore_size)
>>>>> + return TRUE;
>>>>> +
>>>>> if (access(path, F_OK) != 0)
>>>>> return TRUE; /* File does not exist */
>>>>> if (info->flag_force) {
>>>>> @@ -4622,6 +4644,47 @@ write_and_check_space(int fd, void *buf, size_t buf_size, char *file_name)
>>>>> return TRUE;
>>>>> }
>>>>>
>>>>> +static int
>>>>> +write_buffer_update_size_info(off_t offset, void *buf, size_t buf_size)
>>>>> +{
>>>>> + struct dumpsize_info *dumpsize_info = &info->dumpsize_info;
>>>>> + int blk_end_idx = (offset + buf_size - 1) / dumpsize_info->blksize;
>>>>> + int i;
>>>>> +
>>>>> + /* Need to grow the dumpsize block buffer? */
>>>>> + if (blk_end_idx >= dumpsize_info->block_buff_size) {
>>>>> + int alloc_size = MAX(blk_end_idx - dumpsize_info->block_buff_size, BASE_NUM_BLOCKS);
>>>>> +
>>>>> + dumpsize_info->block_info = realloc(dumpsize_info->block_info,
>>>>> + dumpsize_info->block_buff_size + alloc_size);
>>>>> + if (!dumpsize_info->block_info) {
>>>>> + ERRMSG("Not enough memory\n");
>>>>> + return FALSE;
>>>>> + }
>>>>> +
>>>>> + memset(dumpsize_info->block_info + dumpsize_info->block_buff_size,
>>>>> + 0, alloc_size);
>>>>> + dumpsize_info->block_buff_size += alloc_size;
>>>>> + }
>>>>> +
>>>>> + for (i = 0; i < buf_size; ++i) {
>>>>> + int blk_idx = (offset + i) / dumpsize_info->blksize;
>>>>> +
>>>>> + if (dumpsize_info->block_info[blk_idx]) {
>>>>> + i += dumpsize_info->blksize;
>>>>> + i = i - (i % dumpsize_info->blksize) - 1;
>>>>> + continue;
>>>>> + }
>>>>> +
>>>>> + if (((char *) buf)[i] != 0) {
>>>>> + dumpsize_info->non_hole_blocks++;
>>>>> + dumpsize_info->block_info[blk_idx] = 1;
>>>>> + }
>>>>> + }
>>>>> +
>>>>> + return TRUE;
>>>>> +}
>>>>> +
>>>>> int
>>>>> write_buffer(int fd, off_t offset, void *buf, size_t buf_size, char *file_name)
>>>>> {
>>>>> @@ -4643,6 +4706,8 @@ write_buffer(int fd, off_t offset, void *buf, size_t buf_size, char *file_name)
>>>>> }
>>>>> if (!write_and_check_space(fd, &fdh, sizeof(fdh), file_name))
>>>>> return FALSE;
>>>>> + } else if (info->flag_vmcore_size && fd == info->fd_dumpfile) {
>>>>> + return write_buffer_update_size_info(offset, buf, buf_size);
>
> Why do we need this function? makedumpfile actually writes zero-filled
> pages to the dumpfile with -d 0, and doesn't write them with -d 1.
> So isn't "write_bytes += buf_size" enough? For example, with -d 30,
>
The reason I went with this method was to make an estimate of the number
of blocks actually allocated on the disk (since depending on how the
data written is scattered in the file, there might be a significant
difference between bytes written vs actual size allocated on disk). But
I realize that there is some misunderstanding from my end since written
0 do make block allocation as opposed to not writing at some offset
(skipping the with lseek() ), I would need to fix that.
To highlight the behaviour I'm talking about:
$ dd if=/dev/zero of=./testfile bs=4096 count=1 seek=1
1+0 records in
1+0 records out
4096 bytes (4.1 kB, 4.0 KiB) copied, 0.000302719 s, 13.5 MB/s
$ du -h testfile
4.0K testfile
$ dd if=/dev/zero of=./testfile bs=4096 count=2
2+0 records in
2+0 records out
8192 bytes (8.2 kB, 8.0 KiB) copied, 0.000373002 s, 22.0 MB/s
$ du -h testfile
8.0K testfile
So, do you think it's not worth bothering estimating the number of
blocks allocated an that I should only consider the number of bytes written?
> # makedumpfile --vmcore-size -d30 vmcore
>
> Estimated size to save vmcore is: 147595264 Bytes
> write_bytes: 172782736 Bytes // calculated by "write_bytes += buf_size"
>
> makedumpfile Completed.
> # makedumpfile -d30 vmcore dump.d30
> Copying data : [100.0 %] / eta: 0s
>
> The dumpfile is saved to dump.d30.
>
> makedumpfile Completed.
> # ls -ls dump.d30
> 168740 -rw------- 1 root root 172787864 Oct 16 15:14 dump.d30
>
>
>>>>> } else {
>>>>> if (lseek(fd, offset, SEEK_SET) == failed) {
>>>>> ERRMSG("Can't seek the dump file(%s). %s\n",
>>>>> @@ -9018,6 +9083,12 @@ close_dump_file(void)
>>>>> if (info->flag_flatten)
>>>>> return;
>>>>>
>>>>> + if (info->flag_vmcore_size && info->fd_dumpfile == -1) {
>>>>> + free(info->dumpsize_info.block_info);
>>>>> + info->dumpsize_info.block_info = NULL;
>>>>> + return;
>>>>> + }
>>>>> +
>>>>> if (close(info->fd_dumpfile) < 0)
>>>>> ERRMSG("Can't close the dump file(%s). %s\n",
>>>>> info->name_dumpfile, strerror(errno));
>>>>> @@ -10963,6 +11034,12 @@ check_param_for_creating_dumpfile(int argc, char *argv[])
>>>>> if (info->flag_flatten && info->flag_split)
>>>>> return FALSE;
>>>>>
>>>>> + if (info->flag_flatten && info->flag_vmcore_size)
>>>>> + return FALSE;
>>>>> +
>>>>> + if (info->flag_mem_usage && info->flag_vmcore_size)
>>>>> + return FALSE;
>>>>> +
>>>>> if (info->name_filterconfig && !info->name_vmlinux)
>>>>> return FALSE;
>>>>>
>>>>> @@ -11043,7 +11120,8 @@ check_param_for_creating_dumpfile(int argc, char *argv[])
>>>>> */
>>>>> info->name_memory = argv[optind];
>>>>>
>>>>> - } else if ((argc == optind + 1) && info->flag_mem_usage) {
>>>>> + } else if ((argc == optind + 1) && (info->flag_mem_usage ||
>>>>> + info->flag_vmcore_size)) {
>>>>> /*
>>>>> * Parameter for showing the page number of memory
>>>>> * in different use from.
>>>>> @@ -11423,6 +11501,7 @@ static struct option longopts[] = {
>>>>> {"work-dir", required_argument, NULL, OPT_WORKING_DIR},
>>>>> {"num-threads", required_argument, NULL, OPT_NUM_THREADS},
>>>>> {"check-params", no_argument, NULL, OPT_CHECK_PARAMS},
>>>>> + {"vmcore-size", no_argument, NULL, OPT_VMCORE_SIZE},
>>>>> {0, 0, 0, 0}
>>>>> };
>>>>>
>>>>> @@ -11589,6 +11668,9 @@ main(int argc, char *argv[])
>>>>> info->flag_check_params = TRUE;
>>>>> message_level = DEFAULT_MSG_LEVEL;
>>>>> break;
>>>>> + case OPT_VMCORE_SIZE:
>>>>> + info->flag_vmcore_size = TRUE;
>>>>> + break;
>>>>> case '?':
>>>>> MSG("Commandline parameter is invalid.\n");
>>>>> MSG("Try `makedumpfile --help' for more information.\n");
>>>>> @@ -11598,6 +11680,10 @@ main(int argc, char *argv[])
>>>>> if (flag_debug)
>>>>> message_level |= ML_PRINT_DEBUG_MSG;
>>>>>
>>>>> + if (info->flag_vmcore_size)
>>>>> + /* Suppress progress indicator as dumpfile won't get written */
>>>>> + message_level &= ~ML_PRINT_PROGRESS;
>>>>> +
>>>>> if (info->flag_check_params)
>>>>> /* suppress debugging messages */
>>>>> message_level = DEFAULT_MSG_LEVEL;
>>>>> @@ -11751,7 +11837,11 @@ main(int argc, char *argv[])
>>>>> goto out;
>>>>>
>>>>> MSG("\n");
>>>>> - if (info->flag_split) {
>>>>> +
>>>>> + if (info->flag_vmcore_size) {
>>>>> + MSG("Estimated size to save vmcore is: %lld Bytes\n",
>>>>> + (long long)info->dumpsize_info.non_hole_blocks *
>> info->dumpsize_info.blksize);
>>>>> + } else if (info->flag_split) {
>>>>> MSG("The dumpfiles are saved to ");
>>>>> for (i = 0; i < info->num_dumpfile; i++) {
>>>>> if (i != (info->num_dumpfile - 1))
>>>>> @@ -11808,6 +11898,8 @@ out:
>>>>> free(info->page_buf);
>>>>> if (info->parallel_info != NULL)
>>>>> free(info->parallel_info);
>>>>> + if (info->dumpsize_info.block_info != NULL)
>>>>> + free(info->dumpsize_info.block_info);
>>>>> free(info);
>>>>>
>>>>> if (splitblock) {
>>>>> diff --git a/makedumpfile.h b/makedumpfile.h
>>>>> index 03fb4ce..fd78d5f 100644
>>>>> --- a/makedumpfile.h
>>>>> +++ b/makedumpfile.h
>>>>> @@ -1277,6 +1277,15 @@ struct parallel_info {
>>>>> #endif
>>>>> };
>>>>>
>>>>> +#define BASE_NUM_BLOCKS 50
>>>>> +
>>>>> +struct dumpsize_info {
>>>>> + int blksize;
>>>>> + int block_buff_size;
>>>>> + unsigned char *block_info;
>>>>> + int non_hole_blocks;
>>>>> +};
>>>>> +
>>>>> struct ppc64_vmemmap {
>>>>> unsigned long phys;
>>>>> unsigned long virt;
>>>>> @@ -1321,6 +1330,7 @@ struct DumpInfo {
>>>>> int flag_vmemmap; /* kernel supports vmemmap address space */
>>>>> int flag_excludevm; /* -e - excluding unused vmemmap pages */
>>>>> int flag_use_count; /* _refcount is named _count in struct page */
>>>>> + int flag_vmcore_size; /* estimate the size of the vmcore file instead of
>> creating it */
>>>>> unsigned long vaddr_for_vtop; /* virtual address for debugging */
>>>>> long page_size; /* size of page */
>>>>> long page_shift;
>>>>> @@ -1425,6 +1435,7 @@ struct DumpInfo {
>>>>> int num_dumpfile;
>>>>> struct splitting_info *splitting_info;
>>>>> struct parallel_info *parallel_info;
>>>>> + struct dumpsize_info dumpsize_info;
>>>>>
>>>>> /*
>>>>> * bitmap info:
>>>>> @@ -2364,6 +2375,7 @@ struct elf_prstatus {
>>>>> #define OPT_NUM_THREADS OPT_START+16
>>>>> #define OPT_PARTIAL_DMESG OPT_START+17
>>>>> #define OPT_CHECK_PARAMS OPT_START+18
>>>>> +#define OPT_VMCORE_SIZE OPT_START+19
>>>>>
>>>>> /*
>>>>> * Function Prototype.
>>>>> diff --git a/print_info.c b/print_info.c
>>>>> index e0c38b4..6f5a165 100644
>>>>> --- a/print_info.c
>>>>> +++ b/print_info.c
>>>>> @@ -308,6 +308,10 @@ print_usage(void)
>>>>> MSG(" the crashkernel range, then calculates the page number of different kind per\n");
>>>>> MSG(" vmcoreinfo. So currently /proc/kcore need be specified explicitly.\n");
>>>>> MSG("\n");
>>>>> + MSG(" [--vmcore-size]:\n");
>>>>> + MSG(" This option provides an estimation of the size needed to save VMCORE on disk.\n");
>>>>> + MSG(" This option option cannot be used in combination with -F.\n");
>>>>
>>>> Also not in combination with --mem-usage (as per the code changes above)?
>>>> And may be the options '--mem-usage / -F' also need an update to
>>>> mention they can't be used with --vmcore-size option.
>>>>
>>>
>>> Good point, I'll update those.
>>>
>>>>> + MSG("\n");
>>>>> MSG(" [-D]:\n");
>>>>> MSG(" Print debugging message.\n");
>>>>> MSG("\n");
>>>>> --
>>>>
>>>> I like the idea, but sometimes we use makedumpfile to generate a
>>>> dumpfile in the primary kernel as well. For example:
>>>>
>>>> $ makedumpfile -d 31 -x vmlinux /proc/kcore dumpfile
>>>>
>>>> In such use-cases it is useful to use --vmcore-size and still generate
>>>> the dumpfile (right now the default behaviour is not to generate a
>>>> dumpfile when --vmcore-size is specified). Maybe we need to think more
>>>> on supporting this use-case as well.
>>>>
>>>
>>> The thing is, if you are generating the dumpfile, you can just check the
>>> size of the file created with "du -b" or some other command.
>>
>> I agree, but I just was looking to replace the two 'makedumpfile +
>> du' steps with a single 'makedumpfile --vmcore-size' step.
>>
>>> Overall I don't mind supporting your case as well. Maybe that can depend
>>> on whether a vmcore/dumpfile filename is provided:
>>>
>>> $ makedumpfile -d 31 -x vmlinux /proc/kcore # only estimates the size
>>>
>>> $ makedumpfile -d 31 -x vmlinux /proc/kcore dumpfile # writes the
>>> dumpfile and gives the final size
>>>
>>> Any thought, opinions, suggestions?
>>
>> Let's wait for Kazu's opinion on the same, but I am ok with using a
>> two-step 'makedumpfile + du' approach for now (and later expand
>> --vmcore-size as we encounter more use-cases).
>>
>> @Kazuhito Hagio : What's your opinion on the above?
>
> I would prefer only estimating with the option.
>
> And if the write_bytes method above is usable, it can be shown also
> in report messages when wrote the dumpfile.
>
Let me know your preferred approach considering my comment above and
I'll send out a v2.
Thanks,
--
Julien Thierry
More information about the kexec
mailing list