| Summary: | Regression(r179584): Assertion hit in toResourceLoadPriority() on Yosemite | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Chris Dumez <cdumez> | ||||
| Component: | Page Loading | Assignee: | Chris Dumez <cdumez> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | ap, koivisto | ||||
| Priority: | P2 | ||||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Bug Depends on: | |||||||
| Bug Blocks: | 141186 | ||||||
| Attachments: |
|
||||||
|
Description
Chris Dumez
2015-02-03 21:21:13 PST
Created attachment 246014 [details]
Patch
> The CFNetwork doc (CFURLRequestPriority) does not indicate -1 as being a valid resource load priority value.
I recall that we define custom priority levels and don't use normal CFNetwork ones. Perhaps -1 is one of those?
(In reply to comment #2) > > The CFNetwork doc (CFURLRequestPriority) does not indicate -1 as being a valid resource load priority value. > > I recall that we define custom priority levels and don't use normal > CFNetwork ones. Perhaps -1 is one of those? Where would those be? We always call toResourceLoadPriority(wkGetHTTPRequestPriority()). I looked at the wkGetHTTPRequestPriority() implementation but couldn't find anything custom. Also, when calling wkSetHTTPRequestPriority(), we always use toPlatformRequestPriority() on the priority. I took care in r179584 to remove -1 from that function already so we should never be setting -1 ourselves. Looking at CFURLRequest.h, CFURLRequestPriority is documented to have values from 0 to 2. However, we call WKSetHTTPRequestMaximumPriority to change that, and use values up to 4. This indeed doesn't answer the question of where -1 comes from. Did you check CFNetwork source code? (In reply to comment #5) > Looking at CFURLRequest.h, CFURLRequestPriority is documented to have values > from 0 to 2. However, we call WKSetHTTPRequestMaximumPriority to change > that, and use values up to 4. > > This indeed doesn't answer the question of where -1 comes from. Did you > check CFNetwork source code? Yes, this is what I am currently doing (so far no explanation though). I still think we should land this patch in the mean time though to make the bots happy. We were handling -1 in this function prior to r179584 already. Comment on attachment 246014 [details] Patch Clearing flags on attachment: 246014 Committed r179593: <http://trac.webkit.org/changeset/179593> All reviewed patches have been landed. Closing bug. |