Assertion 'xfered>0' in target.c for remote connection
Commit Message
(Resending, since my previous response lost threading information in the
mailing list)
On 11/2/2017 10:11 PM, Sergio Durigan Junior wrote:
> On Thursday, November 02 2017, pcarroll@codesourcery.com wrote:
>
>> We have a customer who is using a Corelis gdb server to connect to gdb.
>> Occasionally, the gdb server will send a 0-byte block of memory for a read.
>> When this happens, gdb gives an assertion from target.c:
>>
>> internal-error: target_xfer_partial: Assertion `*xfered_len > 0' failed.
>>
>> This problem is almost identical to that fixed in https://sourceware.org/ml/gdb-patches/2014-02/msg00636.html
>>
>> In this case, remote.c needs to be modified to return TARGET_XFER_EOF instead of TARGET_XFER_OK or TARGET_XFER_UNAVAILABLE when 0 bytes are transferred.
> Thanks for the patch. A few comments below. BTW, your mail client
> seems to have messed up with the formatting of the patch, removing TABs
> and replacing them with whitespaces. Could you please send the patch
> again, but using a client that handles patches better?
Yep, resending here. I don't think I've sent email to the mailing lists
for years, so I needed to use a different mail client.
>> The proposed fix would be:
>>
>> diff -rup fsf/gdb/ChangeLog fix/gdb/ChangeLog
>> --- fsf/gdb/ChangeLog 2017-11-02 16:13:19.188615000 -0500
>> +++ fix/gdb/ChangeLog 2017-11-02 16:13:21.626754500 -0500
>> @@ -1,3 +1,10 @@
>> +2017-11-02 Paul Carroll <pcarroll@codesourcery.com>
>> +
>> + PR gdb/22388
>> + * remote.c (remote_write_bytes_aux, remote_read_bytes_1,
>> + remote_read_bytes, remote_write_qxfer, remote_xfer_partial):
>> + Return TARGET_XFER_EOF if size of returned data is 0.
>> +
> The ChangeLog entries need to be formatted with TABs instead of
> whitespaces. But the text looks great.
Taken care of in the new patch
>> 2017-11-02 Yao Qi <yao.qi@linaro.org>
>>
>> * frame.c (do_frame_register_read): Remove aspace.
>> diff -rup fsf/gdb/remote.c fix/gdb/remote.c
>> --- fsf/gdb/remote.c 2017-11-02 16:06:15.979408800 -0500
>> +++ fix/gdb/remote.c 2017-11-02 15:24:35.536391700 -0500
>> @@ -8264,7 +8264,7 @@ remote_write_bytes_aux (const char *head
>> /* Return UNITS_WRITTEN, not TODO_UNITS, in case escape chars caused us to
>> send fewer units than we'd planned. */
>> *xfered_len_units = (ULONGEST) units_written;
>> - return TARGET_XFER_OK;
>> + return (*xfered_len_units != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
>> }
> This one looks correct to me.
>
>> /* Write memory data directly to the remote machine.
>> @@ -8358,7 +8358,7 @@ remote_read_bytes_1 (CORE_ADDR memaddr,
>> decoded_bytes = hex2bin (p, myaddr, todo_units * unit_size);
>> /* Return what we have. Let higher layers handle partial reads. */
>> *xfered_len_units = (ULONGEST) (decoded_bytes / unit_size);
>> - return TARGET_XFER_OK;
>> + return (*xfered_len_units != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
>> }
> Likewise.
>
>> /* Using the set of read-only target sections of remote, read live
>> @@ -8455,13 +8455,14 @@ remote_read_bytes (struct target_ops *op
>> res = remote_xfer_live_readonly_partial (ops, myaddr, memaddr,
>> len, unit_size, xfered_len);
>> if (res == TARGET_XFER_OK)
>> - return TARGET_XFER_OK;
>> + return (*xfered_len != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
> remote_xfer_live_readonly_partial can only return TARGET_XFER_EOF or
> what remote_read_bytes_1 return (which already takes care of the EOF
> part), so I don't think you need this part here, although keeping it
> wouldn't hurt.
I got rid of this in the new patch, since it doesn't add anything, as
you noted.
>> else
>> {
>> /* No use trying further, we know some memory starting
>> at MEMADDR isn't available. */
>> *xfered_len = len;
>> - return TARGET_XFER_UNAVAILABLE;
>> + return (*xfered_len != 0) ?
>> + TARGET_XFER_UNAVAILABLE : TARGET_XFER_EOF;
>> }
>> }
>>
>> @@ -10386,7 +10387,7 @@ remote_write_qxfer (struct target_ops *o
>> unpack_varlen_hex (rs->buf, &n);
>>
>> *xfered_len = n;
>> - return TARGET_XFER_OK;
>> + return (*xfered_len != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
>> }
>>
>> /* Read OBJECT_NAME/ANNEX from the remote target using a qXfer packet.
>> @@ -10687,7 +10688,7 @@ remote_xfer_partial (struct target_ops *
>> strcpy ((char *) readbuf, rs->buf);
>>
>> *xfered_len = strlen ((char *) readbuf);
>> - return TARGET_XFER_OK;
>> + return (*xfered_len != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
>> }
>>
>> /* Implementation of to_get_memory_xfer_limit. */
> The rest looks sane to me, but I'm not a global maintainer and cannot
> approve your patch. Meanwhile I'd recommend resending the patch to fix
> the formatting errors.
>
> Cheers,
>
Comments
On 2017-11-14 10:02 AM, Paul Carroll wrote:
>>> We have a customer who is using a Corelis gdb server to connect to gdb.
>>> Occasionally, the gdb server will send a 0-byte block of memory for a read.
>>> When this happens, gdb gives an assertion from target.c:
>>>
>>> internal-error: target_xfer_partial: Assertion `*xfered_len > 0' failed.
>>>
>>> This problem is almost identical to that fixed in https://sourceware.org/ml/gdb-patches/2014-02/msg00636.html
>>>
>>> In this case, remote.c needs to be modified to return TARGET_XFER_EOF instead of TARGET_XFER_OK or TARGET_XFER_UNAVAILABLE when 0 bytes are transferred.
The patch look good to me, given that all other implementations do this.
It is small enough that it doesn't require a copyright assignment I think.
I see you have contributed to binutils in the past. Do you already have
push access to the binutils-gdb repo? If not we can push it for you, or
we can get you an account if you plan on contributing regularly.
Thanks,
Simon
On Tue, Nov 14, 2017 at 9:38 PM, Simon Marchi <simon.marchi@ericsson.com> wrote:
>
> It is small enough that it doesn't require a copyright assignment I think.
>
Mentor Graphics had the copyright assignment for all @codesourcery.com
email addresses.
On 11/14/2017 4:38 PM, Simon Marchi wrote:
> On 2017-11-14 10:02 AM, Paul Carroll wrote:
>>>> We have a customer who is using a Corelis gdb server to connect to gdb.
>>>> Occasionally, the gdb server will send a 0-byte block of memory for a read.
>>>> When this happens, gdb gives an assertion from target.c:
>>>>
>>>> internal-error: target_xfer_partial: Assertion `*xfered_len > 0' failed.
>>>>
>>>> This problem is almost identical to that fixed in https://sourceware.org/ml/gdb-patches/2014-02/msg00636.html
>>>>
>>>> In this case, remote.c needs to be modified to return TARGET_XFER_EOF instead of TARGET_XFER_OK or TARGET_XFER_UNAVAILABLE when 0 bytes are transferred.
> The patch look good to me, given that all other implementations do this.
> It is small enough that it doesn't require a copyright assignment I think.
> I see you have contributed to binutils in the past. Do you already have
> push access to the binutils-gdb repo? If not we can push it for you, or
> we can get you an account if you plan on contributing regularly.
>
Please feel free to push the patch for me.
Hard to say how regularly I will be patching things.
The copyright assignment, as Yao noted, is covered by Mentor Graphics
(and Siemens).
On 2017-11-14 05:30 PM, Paul Carroll wrote:
> On 11/14/2017 4:38 PM, Simon Marchi wrote:
>> On 2017-11-14 10:02 AM, Paul Carroll wrote:
>>>>> We have a customer who is using a Corelis gdb server to connect to gdb.
>>>>> Occasionally, the gdb server will send a 0-byte block of memory for a read.
>>>>> When this happens, gdb gives an assertion from target.c:
>>>>>
>>>>> internal-error: target_xfer_partial: Assertion `*xfered_len > 0' failed.
>>>>>
>>>>> This problem is almost identical to that fixed in https://sourceware.org/ml/gdb-patches/2014-02/msg00636.html
>>>>>
>>>>> In this case, remote.c needs to be modified to return TARGET_XFER_EOF instead of TARGET_XFER_OK or TARGET_XFER_UNAVAILABLE when 0 bytes are transferred.
>> The patch look good to me, given that all other implementations do this.
>> It is small enough that it doesn't require a copyright assignment I think.
>> I see you have contributed to binutils in the past. Do you already have
>> push access to the binutils-gdb repo? If not we can push it for you, or
>> we can get you an account if you plan on contributing regularly.
>>
> Please feel free to push the patch for me.
> Hard to say how regularly I will be patching things.
> The copyright assignment, as Yao noted, is covered by Mentor Graphics (and Siemens).
It is pushed, thanks again for the patch.
Simon
@@ -1,3 +1,10 @@
+2017-11-02 Paul Carroll <pcarroll@codesourcery.com>
+
+ PR gdb/22388
+ * remote.c (remote_write_bytes_aux, remote_read_bytes_1,
+ remote_read_bytes, remote_write_qxfer, remote_xfer_partial):
+ Return TARGET_XFER_EOF if size of returned data is 0.
+
2017-11-02 Yao Qi <yao.qi@linaro.org>
* frame.c (do_frame_register_read): Remove aspace.
@@ -8264,7 +8264,7 @@ remote_write_bytes_aux (const char *head
/* Return UNITS_WRITTEN, not TODO_UNITS, in case escape chars caused us to
send fewer units than we'd planned. */
*xfered_len_units = (ULONGEST) units_written;
- return TARGET_XFER_OK;
+ return (*xfered_len_units != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
}
/* Write memory data directly to the remote machine.
@@ -8358,7 +8358,7 @@ remote_read_bytes_1 (CORE_ADDR memaddr,
decoded_bytes = hex2bin (p, myaddr, todo_units * unit_size);
/* Return what we have. Let higher layers handle partial reads. */
*xfered_len_units = (ULONGEST) (decoded_bytes / unit_size);
- return TARGET_XFER_OK;
+ return (*xfered_len_units != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
}
/* Using the set of read-only target sections of remote, read live
@@ -8461,7 +8461,8 @@ remote_read_bytes (struct target_ops *op
/* No use trying further, we know some memory starting
at MEMADDR isn't available. */
*xfered_len = len;
- return TARGET_XFER_UNAVAILABLE;
+ return (*xfered_len != 0) ?
+ TARGET_XFER_UNAVAILABLE : TARGET_XFER_EOF;
}
}
@@ -10386,7 +10387,7 @@ remote_write_qxfer (struct target_ops *o
unpack_varlen_hex (rs->buf, &n);
*xfered_len = n;
- return TARGET_XFER_OK;
+ return (*xfered_len != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
}
/* Read OBJECT_NAME/ANNEX from the remote target using a qXfer packet.
@@ -10687,7 +10688,7 @@ remote_xfer_partial (struct target_ops *
strcpy ((char *) readbuf, rs->buf);
*xfered_len = strlen ((char *) readbuf);
- return TARGET_XFER_OK;
+ return (*xfered_len != 0) ? TARGET_XFER_OK : TARGET_XFER_EOF;
}
/* Implementation of to_get_memory_xfer_limit. */