PDF documents loaded into JIRA issues are sent to the client browsers using a mime type of text/html instead of application/pdf. Because Internet Explorer mostly ignores mime-types, it works there but not in Firefox (you have to download it then open it).