[v3,1/1] dump.c: allow fd_write_vmcore to return errno on failure

From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>

fd_write_vmcore can fail to execute for a lot of reasons that can be
retrieved by errno, but it only returns -1. This makes difficult for
the caller to know what happened and only a generic error message is
propagated back to the user. This is an example using dump-guest-memory:

(qemu) dump-guest-memory /home/yasmin/mnt/test.dump
dump: failed to save memory

All callers of fd_write_vmcore of dump.c does error handling via
error_setg(), so at first it seems feasible to add the Error pointer as
an argument of fd_write_vmcore. This proved to be more complex than it
first looked. fd_write_vmcore is used by write_elf64_notes and
write_elf32_notes as a WriteCoreDumpFunction prototype. WriteCoreDumpFunction
is declared in include/qom/cpu.h and is used all around the code. This
leaves us with few alternatives:

- change the WriteCoreDumpFunction prototype to include an error pointer.
This would require to change all functions that implements this prototype
to also receive an Error pointer;

- change both write_elf64_notes and write_elf32_notes to no use the
WriteCoreDumpFunction. These functions use not only fd_write_vmcore
but also buf_write_note, so this would require to change buf_write_note
to handle an Error pointer. Considerable easier than the alternative
above, but it's still a lot of code just for the benefit of the callers
of fd_write_vmcore.

This patch presents an easier solution that benefits all fd_write_vmcore
callers:

- instead of returning -1 on error, return -errno. All existing callers
already checks for ret < 0 so there is no need to change the caller's
logic too much. This also allows the retrieval of the errno.

- all callers were updated to use error_setg_errno instead of just
errno_setg. Now that fd_write_vmcore can return an errno, let's update
all callers so they can benefit from a more detailed error message.

This is the same dump-guest-memory example with this patch applied:

(qemu) dump-guest-memory /home/yasmin/mnt/test.dump
dump: failed to save memory: No space left on device
(qemu)

This example illustrates an error of fd_write_vmcore when called
from write_data. All other callers will benefit from better
error messages as well.

Reported-by: yilzhang@redhat.com
Cc: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
Signed-off-by: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
---
 dump.c | 23 ++++++++++++++---------
 1 file changed, 14 insertions(+), 9 deletions(-)

Hi

On Mon, Feb 12, 2018 at 3:25 PM, Daniel Henrique Barboza
<danielhb@linux.vnet.ibm.com> wrote:
> From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
>
> fd_write_vmcore can fail to execute for a lot of reasons that can be
> retrieved by errno, but it only returns -1. This makes difficult for
> the caller to know what happened and only a generic error message is
> propagated back to the user. This is an example using dump-guest-memory:
>
> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
> dump: failed to save memory
>
> All callers of fd_write_vmcore of dump.c does error handling via
> error_setg(), so at first it seems feasible to add the Error pointer as
> an argument of fd_write_vmcore. This proved to be more complex than it
> first looked. fd_write_vmcore is used by write_elf64_notes and
> write_elf32_notes as a WriteCoreDumpFunction prototype. WriteCoreDumpFunction
> is declared in include/qom/cpu.h and is used all around the code. This
> leaves us with few alternatives:
>
> - change the WriteCoreDumpFunction prototype to include an error pointer.
> This would require to change all functions that implements this prototype
> to also receive an Error pointer;
>
> - change both write_elf64_notes and write_elf32_notes to no use the
> WriteCoreDumpFunction. These functions use not only fd_write_vmcore
> but also buf_write_note, so this would require to change buf_write_note
> to handle an Error pointer. Considerable easier than the alternative
> above, but it's still a lot of code just for the benefit of the callers
> of fd_write_vmcore.
>
> This patch presents an easier solution that benefits all fd_write_vmcore
> callers:
>
> - instead of returning -1 on error, return -errno. All existing callers
> already checks for ret < 0 so there is no need to change the caller's
> logic too much. This also allows the retrieval of the errno.
>
> - all callers were updated to use error_setg_errno instead of just
> errno_setg. Now that fd_write_vmcore can return an errno, let's update
> all callers so they can benefit from a more detailed error message.
>
> This is the same dump-guest-memory example with this patch applied:
>
> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
> dump: failed to save memory: No space left on device
> (qemu)
>
> This example illustrates an error of fd_write_vmcore when called
> from write_data. All other callers will benefit from better
> error messages as well.
>
> Reported-by: yilzhang@redhat.com
> Cc: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
> Signed-off-by: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
> Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>

lgtm,

Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>


> ---
>  dump.c | 23 ++++++++++++++---------
>  1 file changed, 14 insertions(+), 9 deletions(-)
>
> diff --git a/dump.c b/dump.c
> index 7b13baa413..171ff8a3b8 100644
> --- a/dump.c
> +++ b/dump.c
> @@ -107,7 +107,7 @@ static int fd_write_vmcore(const void *buf, size_t size, void *opaque)
>
>      written_size = qemu_write_full(s->fd, buf, size);
>      if (written_size != size) {
> -        return -1;
> +        return -errno;
>      }
>
>      return 0;
> @@ -140,7 +140,7 @@ static void write_elf64_header(DumpState *s, Error **errp)
>
>      ret = fd_write_vmcore(&elf_header, sizeof(elf_header), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write elf header");
> +        error_setg_errno(errp, -ret, "dump: failed to write elf header");
>      }
>  }
>
> @@ -171,7 +171,7 @@ static void write_elf32_header(DumpState *s, Error **errp)
>
>      ret = fd_write_vmcore(&elf_header, sizeof(elf_header), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write elf header");
> +        error_setg_errno(errp, -ret, "dump: failed to write elf header");
>      }
>  }
>
> @@ -194,7 +194,8 @@ static void write_elf64_load(DumpState *s, MemoryMapping *memory_mapping,
>
>      ret = fd_write_vmcore(&phdr, sizeof(Elf64_Phdr), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write program header table");
> +        error_setg_errno(errp, -ret,
> +                         "dump: failed to write program header table");
>      }
>  }
>
> @@ -217,7 +218,8 @@ static void write_elf32_load(DumpState *s, MemoryMapping *memory_mapping,
>
>      ret = fd_write_vmcore(&phdr, sizeof(Elf32_Phdr), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write program header table");
> +        error_setg_errno(errp, -ret,
> +                         "dump: failed to write program header table");
>      }
>  }
>
> @@ -237,7 +239,8 @@ static void write_elf64_note(DumpState *s, Error **errp)
>
>      ret = fd_write_vmcore(&phdr, sizeof(Elf64_Phdr), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write program header table");
> +        error_setg_errno(errp, -ret,
> +                         "dump: failed to write program header table");
>      }
>  }
>
> @@ -302,7 +305,8 @@ static void write_elf32_note(DumpState *s, Error **errp)
>
>      ret = fd_write_vmcore(&phdr, sizeof(Elf32_Phdr), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write program header table");
> +        error_setg_errno(errp, -ret,
> +                         "dump: failed to write program header table");
>      }
>  }
>
> @@ -355,7 +359,8 @@ static void write_elf_section(DumpState *s, int type, Error **errp)
>
>      ret = fd_write_vmcore(&shdr, shdr_size, s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write section header table");
> +        error_setg_errno(errp, -ret,
> +                         "dump: failed to write section header table");
>      }
>  }
>
> @@ -365,7 +370,7 @@ static void write_data(DumpState *s, void *buf, int length, Error **errp)
>
>      ret = fd_write_vmcore(buf, length, s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to save memory");
> +        error_setg_errno(errp, -ret, "dump: failed to save memory");
>      } else {
>          s->written_size += length;
>      }
> --
> 2.14.3
>

On 02/12/2018 12:25 PM, Daniel Henrique Barboza wrote:
> From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
> 
> fd_write_vmcore can fail to execute for a lot of reasons that can be
> retrieved by errno, but it only returns -1. This makes difficult for
> the caller to know what happened and only a generic error message is
> propagated back to the user. This is an example using dump-guest-memory:
> 
> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
> dump: failed to save memory
> 
> All callers of fd_write_vmcore of dump.c does error handling via
> error_setg(), so at first it seems feasible to add the Error pointer as
> an argument of fd_write_vmcore. This proved to be more complex than it
> first looked. fd_write_vmcore is used by write_elf64_notes and
> write_elf32_notes as a WriteCoreDumpFunction prototype. WriteCoreDumpFunction
> is declared in include/qom/cpu.h and is used all around the code. This
> leaves us with few alternatives:
> 
> - change the WriteCoreDumpFunction prototype to include an error pointer.
> This would require to change all functions that implements this prototype
> to also receive an Error pointer;
> 
> - change both write_elf64_notes and write_elf32_notes to no use the
> WriteCoreDumpFunction. These functions use not only fd_write_vmcore
> but also buf_write_note, so this would require to change buf_write_note
> to handle an Error pointer. Considerable easier than the alternative
> above, but it's still a lot of code just for the benefit of the callers
> of fd_write_vmcore.
> 
> This patch presents an easier solution that benefits all fd_write_vmcore
> callers:
> 
> - instead of returning -1 on error, return -errno. All existing callers
> already checks for ret < 0 so there is no need to change the caller's
> logic too much. This also allows the retrieval of the errno.
> 
> - all callers were updated to use error_setg_errno instead of just
> errno_setg. Now that fd_write_vmcore can return an errno, let's update
> all callers so they can benefit from a more detailed error message.
> 
> This is the same dump-guest-memory example with this patch applied:
> 
> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
> dump: failed to save memory: No space left on device
> (qemu)
> 
> This example illustrates an error of fd_write_vmcore when called
> from write_data. All other callers will benefit from better
> error messages as well.
> 
> Reported-by: yilzhang@redhat.com
> Cc: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
> Signed-off-by: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
> Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
> ---
>  dump.c | 23 ++++++++++++++---------
>  1 file changed, 14 insertions(+), 9 deletions(-)
> 
> diff --git a/dump.c b/dump.c
> index 7b13baa413..171ff8a3b8 100644
> --- a/dump.c
> +++ b/dump.c
> @@ -107,7 +107,7 @@ static int fd_write_vmcore(const void *buf, size_t size, void *opaque)
> 
>      written_size = qemu_write_full(s->fd, buf, size);
>      if (written_size != size) {
> -        return -1;
> +        return -errno;
>      }
> 
>      return 0;
> @@ -140,7 +140,7 @@ static void write_elf64_header(DumpState *s, Error **errp)
> 
>      ret = fd_write_vmcore(&elf_header, sizeof(elf_header), s);
>      if (ret < 0) {
> -        error_setg(errp, "dump: failed to write elf header");
> +        error_setg_errno(errp, -ret, "dump: failed to write elf header");

Do we need -ret passed to error_setg_errno()? fd_write_vmcore() returns
negative errno in case of error.

Cheers
Murilo

On 02/12/2018 12:46 PM, Murilo Opsfelder Araujo wrote:
> On 02/12/2018 12:25 PM, Daniel Henrique Barboza wrote:
>> From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
>>
>> fd_write_vmcore can fail to execute for a lot of reasons that can be
>> retrieved by errno, but it only returns -1. This makes difficult for
>> the caller to know what happened and only a generic error message is
>> propagated back to the user. This is an example using dump-guest-memory:
>>
>> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
>> dump: failed to save memory
>>
>> All callers of fd_write_vmcore of dump.c does error handling via
>> error_setg(), so at first it seems feasible to add the Error pointer as
>> an argument of fd_write_vmcore. This proved to be more complex than it
>> first looked. fd_write_vmcore is used by write_elf64_notes and
>> write_elf32_notes as a WriteCoreDumpFunction prototype. WriteCoreDumpFunction
>> is declared in include/qom/cpu.h and is used all around the code. This
>> leaves us with few alternatives:
>>
>> - change the WriteCoreDumpFunction prototype to include an error pointer.
>> This would require to change all functions that implements this prototype
>> to also receive an Error pointer;
>>
>> - change both write_elf64_notes and write_elf32_notes to no use the
>> WriteCoreDumpFunction. These functions use not only fd_write_vmcore
>> but also buf_write_note, so this would require to change buf_write_note
>> to handle an Error pointer. Considerable easier than the alternative
>> above, but it's still a lot of code just for the benefit of the callers
>> of fd_write_vmcore.
>>
>> This patch presents an easier solution that benefits all fd_write_vmcore
>> callers:
>>
>> - instead of returning -1 on error, return -errno. All existing callers
>> already checks for ret < 0 so there is no need to change the caller's
>> logic too much. This also allows the retrieval of the errno.
>>
>> - all callers were updated to use error_setg_errno instead of just
>> errno_setg. Now that fd_write_vmcore can return an errno, let's update
>> all callers so they can benefit from a more detailed error message.
>>
>> This is the same dump-guest-memory example with this patch applied:
>>
>> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
>> dump: failed to save memory: No space left on device
>> (qemu)
>>
>> This example illustrates an error of fd_write_vmcore when called
>> from write_data. All other callers will benefit from better
>> error messages as well.
>>
>> Reported-by: yilzhang@redhat.com
>> Cc: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
>> Signed-off-by: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
>> Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
>> ---
>>   dump.c | 23 ++++++++++++++---------
>>   1 file changed, 14 insertions(+), 9 deletions(-)
>>
>> diff --git a/dump.c b/dump.c
>> index 7b13baa413..171ff8a3b8 100644
>> --- a/dump.c
>> +++ b/dump.c
>> @@ -107,7 +107,7 @@ static int fd_write_vmcore(const void *buf, size_t size, void *opaque)
>>
>>       written_size = qemu_write_full(s->fd, buf, size);
>>       if (written_size != size) {
>> -        return -1;
>> +        return -errno;
>>       }
>>
>>       return 0;
>> @@ -140,7 +140,7 @@ static void write_elf64_header(DumpState *s, Error **errp)
>>
>>       ret = fd_write_vmcore(&elf_header, sizeof(elf_header), s);
>>       if (ret < 0) {
>> -        error_setg(errp, "dump: failed to write elf header");
>> +        error_setg_errno(errp, -ret, "dump: failed to write elf header");
> Do we need -ret passed to error_setg_errno()? fd_write_vmcore() returns
> negative errno in case of error.

I am passing -ret here to provide a positive errno value to 
error_setg_errno. I
am not sure if os_error can be negative and didn't want to risk 
strerror() breaking.



Daniel


>
> Cheers
> Murilo

On 02/12/2018 08:46 AM, Murilo Opsfelder Araujo wrote:
> On 02/12/2018 12:25 PM, Daniel Henrique Barboza wrote:
>> From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
>>
>> fd_write_vmcore can fail to execute for a lot of reasons that can be
>> retrieved by errno, but it only returns -1. This makes difficult for
>> the caller to know what happened and only a generic error message is
>> propagated back to the user. This is an example using dump-guest-memory:
>>

>> +++ b/dump.c
>> @@ -107,7 +107,7 @@ static int fd_write_vmcore(const void *buf, size_t size, void *opaque)
>>
>>       written_size = qemu_write_full(s->fd, buf, size);
>>       if (written_size != size) {
>> -        return -1;
>> +        return -errno;
>>       }
>>
>>       return 0;
>> @@ -140,7 +140,7 @@ static void write_elf64_header(DumpState *s, Error **errp)
>>
>>       ret = fd_write_vmcore(&elf_header, sizeof(elf_header), s);
>>       if (ret < 0) {
>> -        error_setg(errp, "dump: failed to write elf header");
>> +        error_setg_errno(errp, -ret, "dump: failed to write elf header");
> 
> Do we need -ret passed to error_setg_errno()? fd_write_vmcore() returns
> negative errno in case of error.

Yes, this usage is correct.  error_setg_errno() takes a positive errno 
value (using strerror, which only decodes positive values into useful 
strings); but we typically return negative errno values (as was 
correctly done in fd_write_vmcore), so the extra layer of negation here 
is needed.

On 02/12/2018 08:25 AM, Daniel Henrique Barboza wrote:
> From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
> 
> fd_write_vmcore can fail to execute for a lot of reasons that can be
> retrieved by errno, but it only returns -1. This makes difficult for
> the caller to know what happened and only a generic error message is
> propagated back to the user. This is an example using dump-guest-memory:
> 
> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
> dump: failed to save memory
> 
...
> This is the same dump-guest-memory example with this patch applied:
> 
> (qemu) dump-guest-memory /home/yasmin/mnt/test.dump
> dump: failed to save memory: No space left on device
> (qemu)
> 
> This example illustrates an error of fd_write_vmcore when called
> from write_data. All other callers will benefit from better
> error messages as well.
> 
> Reported-by: yilzhang@redhat.com
> Cc: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
> Signed-off-by: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
> Signed-off-by: Daniel Henrique Barboza <danielhb@linux.vnet.ibm.com>
> ---
>   dump.c | 23 ++++++++++++++---------
>   1 file changed, 14 insertions(+), 9 deletions(-)
> 

Reviewed-by: Eric Blake <eblake@redhat.com>

On 02/12/2018 03:31 PM, Eric Blake wrote:
> On 02/12/2018 08:46 AM, Murilo Opsfelder Araujo wrote:
>> On 02/12/2018 12:25 PM, Daniel Henrique Barboza wrote:
>>> From: Yasmin Beatriz <yasmins@linux.vnet.ibm.com>
>>>
>>> fd_write_vmcore can fail to execute for a lot of reasons that can be
>>> retrieved by errno, but it only returns -1. This makes difficult for
>>> the caller to know what happened and only a generic error message is
>>> propagated back to the user. This is an example using dump-guest-memory:
>>>
> 
>>> +++ b/dump.c
>>> @@ -107,7 +107,7 @@ static int fd_write_vmcore(const void *buf,
>>> size_t size, void *opaque)
>>>
>>>       written_size = qemu_write_full(s->fd, buf, size);
>>>       if (written_size != size) {
>>> -        return -1;
>>> +        return -errno;
>>>       }
>>>
>>>       return 0;
>>> @@ -140,7 +140,7 @@ static void write_elf64_header(DumpState *s,
>>> Error **errp)
>>>
>>>       ret = fd_write_vmcore(&elf_header, sizeof(elf_header), s);
>>>       if (ret < 0) {
>>> -        error_setg(errp, "dump: failed to write elf header");
>>> +        error_setg_errno(errp, -ret, "dump: failed to write elf
>>> header");
>>
>> Do we need -ret passed to error_setg_errno()? fd_write_vmcore() returns
>> negative errno in case of error.
> 
> Yes, this usage is correct.  error_setg_errno() takes a positive errno
> value (using strerror, which only decodes positive values into useful
> strings); but we typically return negative errno values (as was
> correctly done in fd_write_vmcore), so the extra layer of negation here
> is needed.
> 

For some reason I assumed "non-zero" in the error_setg_errno()
description as negative.

Thanks Daniel and Eric.