Skip to contentSkip to conversation

http: return error on unhandled HTTP error responses

PATCH In review Olaf Hering 2 participants 6 messages Mar 17, 2020 – Apr 6, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 http: return error on unhandled HTTP error responses

A http transfer will hang if an unhandled error is returned.
The error branch returns the value zero, which is not expected by the caller.

---
 grub-core/net/http.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
grub-core/net/http.c +1 -1
index b616cf40b..9d92a4905 100644
@@ -125,7 +125,7 @@ parse_line (grub_file_t file, http_data_t data, char *ptr, grub_size_t len)
 	     valid answers like 403 will trigger this very generic message.  */
 	  data->errmsg = grub_xasprintf (_("unsupported HTTP error %d: %s"),
 					 code, ptr);
-	  return GRUB_ERR_NONE;
+	  return GRUB_ERR_FILE_READ_ERROR;
 	}
       data->first_line_recv = 1;
       return GRUB_ERR_NONE;
#1 lore raw
Quoted 13 lines · Olaf Hering

A http transfer will hang if an unhandled error is returned.
The error branch returns the value zero, which is not expected by the caller.

---
 grub-core/net/http.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
grub-core/net/http.c +0 -0
index b616cf40b..9d92a4905 100644

grub-core/net/http.c parse_line (grub_file_t file, http_data_t data, char *ptr, grub_size_t len) · line 128

Quoted 5 lines
grub-core/net/http.c +1 -1
@@ -125,7 +125,7 @@ parse_line (grub_file_t file, http_data_t data, char *ptr, grub_size_t len)
 	     valid answers like 403 will trigger this very generic message.  */
 	  data->errmsg = grub_xasprintf (_("unsupported HTTP error %d: %s"),
 					 code, ptr);
-	  return GRUB_ERR_NONE;
+	  return GRUB_ERR_FILE_READ_ERROR;

Should not we do the same for 404, file not found, a few lines above?

Daniel

#2 lore raw

Am Wed, 25 Mar 2020 19:55:47 +0100
schrieb Daniel Kiper <[email protected]>:

Quoted 1 line

Should not we do the same for 404, file not found, a few lines above?

Maybe. For some reason a 404 returns quickly, while a 400 will request the file 4 times. With this patch there is still some delay, but the request is sent just once. I wonder what the author had in mind, where the error/state is actually supposed to be checked.

Olaf

#3 lore raw
Quoted 9 lines · Olaf Hering

Am Wed, 25 Mar 2020 19:55:47 +0100
schrieb Daniel Kiper <[email protected]>:

Quoted 1 line

Should not we do the same for 404, file not found, a few lines above?

Maybe. For some reason a 404 returns quickly, while a 400 will request the file 4 times. With this patch there is still some delay, but the request is sent just once. I wonder what the author had in mind, where the error/state is actually supposed to be checked.

May I ask you to do the change for 404 and test it too? If it works please post a new patch.

Daniel

#4 lore raw
Quoted 6 lines · Daniel Kiper
Quoted 5 lines · Olaf Hering

Am Wed, 25 Mar 2020 19:55:47 +0100
schrieb Daniel Kiper <[email protected]>:

Quoted 1 line

Should not we do the same for 404, file not found, a few lines above?

Quoted 7 lines
Quoted 4 lines

Maybe. For some reason a 404 returns quickly, while a 400 will request the file 4 times. With this patch there is still some delay, but the request is sent just once. I wonder what the author had in mind, where the error/state is actually supposed to be checked.

May I ask you to do the change for 404 and test it too? If it works please post a new patch.

Ping?

Daniel

#5 lore raw
In reply to Daniel Kiper

Am Thu, 26 Mar 2020 20:36:47 +0100
schrieb Daniel Kiper <[email protected]>:

Quoted 2 lines

May I ask you to do the change for 404 and test it too? If it works please post a new patch.

I think "http" needs more surgery. It happens to work despite the (apparently) bogus error handling. But I do not have time to work on this now.

Olaf