From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 40868 invoked by alias); 25 Oct 2018 23:59:08 -0000 Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org Received: (qmail 40840 invoked by uid 89); 25 Oct 2018 23:59:05 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-2.0 required=5.0 tests=AWL,BAYES_00,RCVD_IN_DNSWL_NONE,SPF_PASS autolearn=ham version=3.3.2 spammy=Hx-languages-length:783 X-HELO: mail-vk1-f195.google.com Received: from mail-vk1-f195.google.com (HELO mail-vk1-f195.google.com) (209.85.221.195) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Thu, 25 Oct 2018 23:59:04 +0000 Received: by mail-vk1-f195.google.com with SMTP id l186so2661892vke.0 for ; Thu, 25 Oct 2018 16:59:04 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sifive.com; s=google; h=mime-version:from:date:message-id:subject:to:cc; bh=w5d/pK6HMbfpmoff+0MPlwM32K2Wsd/UgvFKKO2usig=; b=MPe0cxjgFEyIO036Vaam9jpvMks2lcS2E8DdQ6L3i8BJVtkpR3oiIdFVBSAkUwOI4L 4UoGQLpwOAfJnYG7wWBLbPQBfKBKvHusvM1AL6DS+WofXVUmaMxjCqSlwhtNC83LeAZZ ufObr6ANIt0BGjk9vc4OQIDIPV9yBZtBkc0YP2R51ZAtRlUj15pa+Ka/Ekgo7UW1YBrY 2zP2x3eJnXSlj0AvLe4tSXB8rh3K9DZiTfuG0RoG4+A3HIthPu0P81WnzLuaBcrI9YpN Lz+BB83wpD79GixDDbbeTN1sAOGJlfGf0FonIIcn0itYPvI8t8V58Xd6vNHsTscQR99J WWDQ== MIME-Version: 1.0 From: Jim Wilson Date: Thu, 25 Oct 2018 23:59:00 -0000 Message-ID: Subject: [PATCH 0/2] RISC-V: Linux signal frame support. To: gdb-patches@sourceware.org Cc: Jim Wilson Content-Type: text/plain; charset="UTF-8" X-SW-Source: 2018-10/txt/msg00604.txt.bz2 This adds initial support for unwinding through signal trampolines. I've tested it by hand and can see correct register values for the frame where the signal occurred. I've tested it with the gdb testsuite on an unleashed board with fc29 and a patched kernel, and I get 37 more passes and 35 fewer failures. The result isn't as good as I hoped, as I'm getting a lot of failures from the fact that stepi into handler doesn't work, but that apparently is not expected to work with software single step which is all I have implemented so far. This also fixes some duplicate comments the way that Simon Marchi suggested back in August, by moving them to a .h file and modifying the .c file to refer to the comments in the .h file. Jim